配置提交

# Conflicts:
#	preview.py

.idea/huojv.iml

@@ -2,7 +2,7 @@
 <module type="PYTHON_MODULE" version="4">
   <component name="NewModuleRootManager">
     <content url="file://$MODULE_DIR$" />
-    <orderEntry type="jdk" jdkName="dnf" jdkType="Python SDK" />
+    <orderEntry type="jdk" jdkName="Python 3.12 virtualenv at C:\Users\Administrator\Downloads\huojv\huojv\.venv" jdkType="Python SDK" />
     <orderEntry type="sourceFolder" forTests="false" />
   </component>
 </module>

.idea/misc.xml

@@ -3,5 +3,5 @@
   <component name="Black">
     <option name="sdkName" value="D:\CONDA\anaconda3" />
   </component>
-  <component name="ProjectRootManager" version="2" project-jdk-name="dnf" project-jdk-type="Python SDK" />
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.12 virtualenv at C:\Users\Administrator\Downloads\huojv\huojv\.venv" project-jdk-type="Python SDK" />
 </project>

CHANGELOG.md

@@ -0,0 +1,99 @@
+# 更新日志
+
+## 最新更新 - 多组配置系统
+
+### 主要变更
+
+#### 1. 新增文件
+
+- **config.py** - 配置管理器，支持多组配置
+- **gui_config.py** - 可视化配置界面（tkinter）
+- **preview.py** - 采集卡预览窗口系统
+- **launcher.py** - 统一启动器
+- **README.md** - 使用文档
+- **config.json** - 配置文件（自动生成）
+
+#### 2. 修改的文件
+
+- **main.py** - 完全重写，支持配置系统
+  - 从配置文件加载参数
+  - 支持动态读取v值
+  - 移除硬编码配置
+- **utils/mouse.py** - 移除硬编码串口
+  - 新增 `init_mouse_keyboard()` 函数
+  - 支持从配置初始化
+- **utils/get_image.py** - 支持从配置初始化
+  - GetImage改为延迟初始化
+
+### 核心功能
+
+1. **多组配置管理**
+   - 支持创建多个配置组
+   - 每个配置组独立的串口、采集卡、速度设置
+   - 可视化切换活动配置
+
+2. **可视化配置界面**
+   - GUI界面管理所有配置
+   - 实时编辑和保存
+   - 支持添加/删除配置组
+
+3. **采集卡预览系统**
+   - 网格方式预览多个采集卡
+   - 点击放大查看
+   - 实时画面更新
+
+4. **灵活的参数配置**
+   - 串口配置（端口、波特率）
+   - 采集卡配置（索引、分辨率）
+   - 移动速度(v值)配置
+   - 预览窗口配置
+
+### 使用方式
+
+#### 旧版本（已保留 main_old.py）
+```bash
+python main_old.py  # 直接运行，使用硬编码配置
+```
+
+#### 新版本
+```bash
+# 方式1: 使用启动器（推荐）
+python launcher.py
+
+# 方式2: 直接运行
+python main.py  # 读取config.json的active配置
+```
+
+### 配置示例
+
+```json
+{
+    "groups": [
+        {
+            "name": "配置组1",
+            "serial_port": "COM6",
+            "serial_baudrate": 9600,
+            "camera_index": 0,
+            "camera_width": 1920,
+            "camera_height": 1080,
+            "move_velocity": 470,  // ← move_to函数中的v值
+            "active": true
+        }
+    ]
+}
+```
+
+### 兼容性
+
+- ✅ 保留所有原有功能
+- ✅ YOLO模型无需修改
+- ✅ 游戏逻辑完全兼容
+- ⚠️ 旧代码会使用默认配置
+
+### 下一步
+
+如需添加更多配置：
+1. 在 `config.py` 的 `default_config` 中添加新字段
+2. 在 `gui_config.py` 中添加对应的输入框
+3. 在 `main.py` 中读取并使用新配置
+

PREVIEW_CONFIG_GUIDE.md

@@ -0,0 +1,161 @@
+# 预览配置说明
+
+## 📖 预览配置参数说明
+
+预览配置位于配置界面的**"预览配置"**区域，用于控制采集卡预览窗口的显示方式。
+
+### 配置参数
+
+#### 1. **预览宽度** (`preview_width`)
+- **说明**：预览窗口的宽度（像素）
+- **默认值**：1000
+- **建议值**：
+  - 单个采集卡：800-1200
+  - 2个采集卡：1000-1600
+  - 4个采集卡：1200-2000
+- **用途**：控制预览窗口的整体宽度
+
+#### 2. **预览高度** (`preview_height`)
+- **说明**：预览窗口的高度（像素）
+- **默认值**：700
+- **建议值**：
+  - 单个采集卡：600-900
+  - 2个采集卡：700-1000
+  - 4个采集卡：800-1200
+- **用途**：控制预览窗口的整体高度
+
+#### 3. **预览列数** (`preview_columns`)
+- **说明**：预览窗口中每行显示的采集卡数量
+- **默认值**：2
+- **范围**：1-4（推荐）
+- **用途**：决定采集卡预览的网格布局列数
+- **示例**：
+  - 2列：适合2-4个采集卡
+  - 1列：适合单个采集卡（全屏显示）
+  - 3列：适合3个或6个采集卡
+
+#### 4. **预览行数** (`preview_rows`)
+- **说明**：预览窗口中每列显示的采集卡数量
+- **默认值**：2
+- **范围**：1-4（推荐）
+- **用途**：决定采集卡预览的网格布局行数
+- **示例**：
+  - 2行：配合2列可以显示4个采集卡
+  - 1行：适合1-3个采集卡（水平排列）
+  - 3行：适合3个或更多采集卡
+
+### 📐 配置示例
+
+#### 示例1：单个采集卡全屏显示
+```
+预览宽度：1200
+预览高度：900
+预览列数：1
+预览行数：1
+```
+**效果**：单个采集卡占据整个预览窗口
+
+#### 示例2：2个采集卡并排显示
+```
+预览宽度：1600
+预览高度：800
+预览列数：2
+预览行数：1
+```
+**效果**：两个采集卡水平并排显示
+
+#### 示例3：4个采集卡网格显示（2x2）
+```
+预览宽度：1600
+预览高度：1200
+预览列数：2
+预览行数：2
+```
+**效果**：四个采集卡以2x2网格方式显示
+
+#### 示例4：6个采集卡网格显示（3x2）
+```
+预览宽度：1800
+预览高度：1200
+预览列数：3
+预览行数：2
+```
+**效果**：六个采集卡以3x2网格方式显示
+
+### 🎯 配置技巧
+
+1. **根据采集卡数量调整**：
+   - 采集卡数量 = 列数 × 行数
+   - 例如：3个采集卡可以用 3列×1行 或 2列×2行（最后一个位置为空）
+
+2. **根据屏幕尺寸调整**：
+   - 确保预览窗口能够完整显示在你的屏幕上
+   - 如果屏幕较小，可以减小预览宽度和高度
+
+3. **性能考虑**：
+   - 预览窗口越大，CPU占用可能越高
+   - 推荐在1000-1600像素宽度范围内
+
+4. **视觉体验**：
+   - 保持预览窗口的宽高比接近采集卡的宽高比
+   - 例如：采集卡是1920×1080，预览窗口可以是1600×900
+
+### 🔧 如何配置
+
+1. **打开配置界面**：
+   ```bash
+   python gui_config.py
+   ```
+
+2. **找到预览配置区域**：
+   - 在配置界面右侧找到"预览配置"区域
+
+3. **修改参数**：
+   - 在对应的输入框中输入你想要的数值
+   - 点击"保存配置"按钮
+
+4. **测试预览**：
+   - 点击"启动预览"按钮查看效果
+   - 如果效果不理想，可以返回继续调整
+
+### ⚠️ 注意事项
+
+1. **字段名必须正确**：
+   - 配置使用 `preview_width`、`preview_height`、`preview_columns`、`preview_rows`
+   - 不要使用旧版本的 `preview_window_width` 和 `preview_window_height`
+
+2. **数值类型**：
+   - 所有数值必须是整数
+   - 宽度和高度建议在 500-3000 像素之间
+   - 列数和行数建议在 1-4 之间
+
+3. **配置文件位置**：
+   - 配置文件保存在 `config.json` 文件中的 `display` 部分
+   - 修改后会立即生效（需要重启预览窗口）
+
+### 📝 配置文件示例
+
+```json
+{
+  "groups": [...],
+  "display": {
+    "preview_width": 1000,
+    "preview_height": 700,
+    "preview_columns": 2,
+    "preview_rows": 2,
+    "show_preview": true
+  }
+}
+```
+
+### 🚀 快速开始
+
+1. 默认配置（2x2网格，1000x700窗口）适合大多数场景
+2. 如果有多个采集卡，先尝试2列×2行的布局
+3. 如果只有一个采集卡，设置为1列×1行，增大窗口尺寸
+4. 根据实际效果微调，直到满意为止
+
+---
+
+**提示**：预览窗口可以点击任意采集卡画面来放大查看。放大后的窗口会在新窗口中显示，大小固定为1280×720。
+

README.md

@@ -0,0 +1,141 @@
+# 火炬之光自动化脚本
+
+## 功能特点
+
+- ✅ **多配置组管理** - 支持多组配置，可同时管理多套采集卡和串口设置
+- 🖥️ **可视化配置** - 提供图形化界面管理所有配置
+- 👁️ **采集卡预览** - 多窗口网格预览，点击可放大查看
+- ⚙️ **灵活配置** - 所有参数（串口、采集卡、移动速度等）均可配置
+- 🤖 **智能自动化** - 基于YOLO目标检测的游戏自动化
+
+## 快速开始
+
+### 1. 安装依赖
+
+```bash
+pip install opencv-python ultralytics ddddocr ch9329Comm serial PIL numpy
+```
+
+### 2. 启动程序
+
+```bash
+python launcher.py
+```
+
+启动器提供三个选项：
+- **⚙️ 配置管理** - 配置串口、采集卡等参数
+- **👁️ 采集卡预览** - 实时预览采集卡画面
+- **🚀 启动自动化** - 启动游戏自动化脚本
+
+### 3. 配置设置
+
+首次使用请先点击"配置管理"：
+
+1. **添加配置组** - 点击"添加组"创建新配置
+2. **设置参数**：
+   - 串口：选择设备串口号（如 COM6）
+   - 波特率：设置串口波特率（默认9600）
+   - 采集卡索引：选择采集卡设备号
+   - 采集宽度/高度：设置采集分辨率
+   - 移动速度：设置角色移动速度（v值）
+3. **设为活动** - 选择要使用的配置组并设为活动
+4. **保存配置** - 点击"保存配置"
+
+### 4. 预览采集卡
+
+点击"采集卡预览"可以：
+- 网格方式查看所有活动的采集卡
+- 点击任意窗口可放大查看
+- 实时显示采集卡画面
+
+### 5. 启动自动化
+
+配置完成后，点击"启动自动化"开始游戏自动化。
+
+## 配置说明
+
+### 配置组结构
+
+```json
+{
+  "groups": [
+    {
+      "name": "配置组1",
+      "serial_port": "COM6",
+      "serial_baudrate": 9600,
+      "camera_index": 0,
+      "camera_width": 1920,
+      "camera_height": 1080,
+      "move_velocity": 470,
+      "active": false
+    }
+  ],
+  "display": {
+    "preview_window_width": 640,
+    "preview_window_height": 360,
+    "preview_columns": 2,
+    "preview_rows": 2,
+    "show_preview": true
+  }
+}
+```
+
+### 关键参数说明
+
+- **serial_port**: 串口号，如 COM1, COM6
+- **serial_baudrate**: 波特率，通常为 9600
+- **camera_index**: 采集卡索引，0/1/2...
+- **camera_width/height**: 采集分辨率
+- **move_velocity**: 角色移动速度（v值），数值越大移动越快
+
+## 文件结构
+
+```
+huojv/
+├── launcher.py          # 启动器
+├── main.py              # 主程序（自动游戏）
+├── config.py            # 配置管理器
+├── gui_config.py        # 配置GUI界面
+├── preview.py           # 预览窗口
+├── utils/
+│   ├── get_image.py     # 采集卡管理
+│   ├── mouse.py         # 鼠标控制
+│   ├── shizi.py         # OCR识别
+│   └── ...
+├── best.pt              # YOLO模型（战斗）
+├── best0.pt             # YOLO模型（城镇）
+└── config.json          # 配置文件（自动生成）
+```
+
+## 使用流程
+
+1. 运行 `python launcher.py`
+2. 点击"配置管理"设置参数
+3. 点击"采集卡预览"测试采集
+4. 点击"启动自动化"开始游戏
+5. 享受自动化！
+
+## 注意事项
+
+- 确保采集卡和串口设备已正确连接
+- 串口号和采集卡索引需要根据实际情况设置
+- 移动速度(v值)需要根据游戏角色属性调整
+- 配置文件保存在 `config.json`
+
+## 问题排查
+
+### 无法打开串口
+- 检查串口号是否正确
+- 确认串口设备已连接
+- 检查其他程序是否占用串口
+
+### 无法打开采集卡
+- 检查采集卡索引是否正确（0, 1, 2...）
+- 确认采集卡驱动已安装
+- 尝试不同的采集卡索引
+
+### 识别不准确
+- 调整移动速度(v值)
+- 检查采集卡画面是否清晰
+- 确认游戏窗口位置正确
+

config.json

@@ -0,0 +1,43 @@
+{
+    "groups": [
+        {
+            "name": "服务器",
+            "serial_port": "COM17",
+            "serial_baudrate": 9600,
+            "camera_index": 1,
+            "camera_width": 1920,
+            "camera_height": 1080,
+            "move_velocity": 470,
+            "active": true
+        },
+        {
+            "name": "2号机",
+            "serial_port": "COM14",
+            "serial_baudrate": 9600,
+            "camera_index": 2,
+            "camera_width": 1920,
+            "camera_height": 1080,
+            "move_velocity": 360,
+            "active": false
+        },
+        {
+            "name": "3号机",
+            "serial_port": "COM16",
+            "serial_baudrate": 9600,
+            "camera_index": 0,
+            "camera_width": 1920,
+            "camera_height": 1080,
+            "move_velocity": 360,
+            "active": false
+        }
+    ],
+    "display": {
+        "preview_width": 1920,
+        "preview_height": 1080,
+        "preview_columns": 2,
+        "preview_rows": 2,
+        "show_preview": true,
+        "preview_multi_window": false,
+        "preview_use_all_groups": true
+    }
+}

config.py

@@ -0,0 +1,122 @@
+import json
+import os
+
+class ConfigManager:
+    """配置管理器"""
+    def __init__(self, config_file='config.json'):
+        self.config_file = config_file
+        self.default_config = {
+            "groups": [
+                {
+                    "name": "配置组1",
+                    "serial_port": "COM6",
+                    "serial_baudrate": 9600,
+                    "camera_index": 0,
+                    "camera_width": 1920,
+                    "camera_height": 1080,
+                    "move_velocity": 470,
+                    "active": False
+                }
+            ],
+            "display": {
+                "preview_width": 1000,
+                "preview_height": 700,
+                "preview_columns": 2,
+                "preview_rows": 2,
+                "show_preview": True,
+                "preview_multi_window": False,
+                "preview_use_all_groups": True
+            }
+        }
+        self.config = self.load_config()
+    
+    def load_config(self):
+        """加载配置文件"""
+        if os.path.exists(self.config_file):
+            try:
+                with open(self.config_file, 'r', encoding='utf-8') as f:
+                    config = json.load(f)
+                    # 确保有默认结构
+                    if 'groups' not in config:
+                        config['groups'] = self.default_config['groups']
+                    if 'display' not in config:
+                        config['display'] = self.default_config['display']
+                    return config
+            except Exception as e:
+                print(f"加载配置文件失败: {e}")
+                return self.default_config.copy()
+        return self.default_config.copy()
+    
+    def save_config(self):
+        """保存配置文件"""
+        try:
+            import os
+            # 确保目录存在
+            config_dir = os.path.dirname(os.path.abspath(self.config_file))
+            if config_dir and not os.path.exists(config_dir):
+                os.makedirs(config_dir, exist_ok=True)
+            
+            with open(self.config_file, 'w', encoding='utf-8') as f:
+                json.dump(self.config, f, ensure_ascii=False, indent=4)
+            print(f"✅ 配置文件已保存到: {os.path.abspath(self.config_file)}")
+            return True
+        except Exception as e:
+            print(f"❌ 保存配置文件失败: {e}")
+            import traceback
+            traceback.print_exc()
+            return False
+    
+    def add_group(self, name=None):
+        """添加配置组"""
+        if name is None:
+            name = f"配置组{len(self.config['groups']) + 1}"
+        
+        new_group = {
+            "name": name,
+            "serial_port": "COM6",
+            "serial_baudrate": 9600,
+            "camera_index": len(self.config['groups']),
+            "camera_width": 1920,
+            "camera_height": 1080,
+            "move_velocity": 470,
+            "active": False
+        }
+        self.config['groups'].append(new_group)
+        return new_group
+    
+    def delete_group(self, index):
+        """删除配置组"""
+        if 0 <= index < len(self.config['groups']):
+            del self.config['groups'][index]
+            return True
+        return False
+    
+    def update_group(self, index, **kwargs):
+        """更新配置组"""
+        if 0 <= index < len(self.config['groups']):
+            self.config['groups'][index].update(kwargs)
+            return True
+        return False
+    
+    def get_active_group(self):
+        """获取当前活动的配置组"""
+        for group in self.config['groups']:
+            if group.get('active', False):
+                return group
+        return None
+    
+    def set_active_group(self, index):
+        """设置活动的配置组"""
+        for i, group in enumerate(self.config['groups']):
+            group['active'] = (i == index)
+        return True
+    
+    def get_group_by_index(self, index):
+        """根据索引获取配置组"""
+        if 0 <= index < len(self.config['groups']):
+            return self.config['groups'][index]
+        return None
+
+# 全局配置管理器实例
+config_manager = ConfigManager()
+

gui_config.py

@@ -0,0 +1,624 @@
+import tkinter as tk
+from tkinter import ttk, messagebox, simpledialog
+import cv2
+from config import config_manager
+import threading
+import serial.tools.list_ports
+
+class ConfigGUI:
+    """配置GUI界面"""
+    def __init__(self):
+        self.root = tk.Tk()
+        self.root.title("配置管理 - 火炬之光自动化")
+        self.root.geometry("1000x720")
+        self.root.minsize(900, 600)  # 设置最小尺寸
+        
+        self.selected_index = 0
+        self.preview_thread = None  # 添加预览线程引用
+        self.setup_ui()
+        self.load_current_config()
+    
+    def setup_ui(self):
+        """设置UI界面"""
+        # 左侧：配置组列表
+        left_frame = ttk.Frame(self.root, padding="10")
+        left_frame.pack(side=tk.LEFT, fill=tk.BOTH, expand=False)
+        
+        ttk.Label(left_frame, text="配置组列表", font=('Arial', 12, 'bold')).pack()
+        
+        # 配置组列表
+        self.group_listbox = tk.Listbox(left_frame, width=20, height=25)
+        self.group_listbox.pack(fill=tk.BOTH, expand=True, pady=5)
+        self.group_listbox.bind('<<ListboxSelect>>', self.on_group_select)
+        
+        # 按钮
+        btn_frame = ttk.Frame(left_frame)
+        btn_frame.pack(fill=tk.X)
+        
+        ttk.Button(btn_frame, text="添加组", command=self.add_group).pack(fill=tk.X, pady=2)
+        ttk.Button(btn_frame, text="删除组", command=self.delete_group).pack(fill=tk.X, pady=2)
+        ttk.Button(btn_frame, text="设为活动", command=self.set_active).pack(fill=tk.X, pady=2)
+        
+        # 右侧：配置详情
+        right_frame = ttk.Frame(self.root, padding="10")
+        right_frame.pack(side=tk.RIGHT, fill=tk.BOTH, expand=True)
+        
+        ttk.Label(right_frame, text="配置详情", font=('Arial', 12, 'bold')).pack()
+        
+        # 配置项
+        self.config_vars = {}
+        
+        # 基本配置
+        basic_frame = ttk.LabelFrame(right_frame, text="基本配置", padding="10")
+        basic_frame.pack(fill=tk.X, pady=5)
+        
+        self.create_entry(basic_frame, "name", "配置组名称:")
+        # 采集卡索引：使用下拉+扫描
+        cam_row = ttk.Frame(basic_frame)
+        cam_row.pack(fill=tk.X, pady=3)
+        ttk.Label(cam_row, text="采集卡索引:", width=15).pack(side=tk.LEFT)
+        self.camera_index_var = tk.StringVar()
+        self.camera_index_cb = ttk.Combobox(cam_row, textvariable=self.camera_index_var, width=23, state="readonly")
+        self.camera_index_cb.pack(side=tk.LEFT, padx=5)
+        ttk.Button(cam_row, text="扫描采集卡", command=self.scan_cameras).pack(side=tk.LEFT, padx=5)
+        # 将变量放进config_vars统一管理
+        self.config_vars["camera_index"] = self.camera_index_var
+        self.create_entry(basic_frame, "camera_width", "采集宽度:")
+        self.create_entry(basic_frame, "camera_height", "采集高度:")
+        
+        # 串口配置
+        serial_frame = ttk.LabelFrame(right_frame, text="串口配置", padding="10")
+        serial_frame.pack(fill=tk.X, pady=5)
+        
+        # 串口：使用下拉+扫描
+        port_row = ttk.Frame(serial_frame)
+        port_row.pack(fill=tk.X, pady=3)
+        ttk.Label(port_row, text="串口:", width=15).pack(side=tk.LEFT)
+        self.serial_port_var = tk.StringVar()
+        self.serial_port_cb = ttk.Combobox(port_row, textvariable=self.serial_port_var, width=23, state="readonly")
+        self.serial_port_cb.pack(side=tk.LEFT, padx=5)
+        ttk.Button(port_row, text="扫描串口", command=self.scan_ports).pack(side=tk.LEFT, padx=5)
+        # 将变量放进config_vars统一管理
+        self.config_vars["serial_port"] = self.serial_port_var
+        
+        self.create_entry(serial_frame, "serial_baudrate", "波特率:")
+        
+        # 游戏配置
+        game_frame = ttk.LabelFrame(right_frame, text="游戏配置", padding="10")
+        game_frame.pack(fill=tk.X, pady=5)
+        
+        self.create_entry(game_frame, "move_velocity", "移动速度(v值):")
+        
+        # 预览配置
+        preview_frame = ttk.LabelFrame(right_frame, text="预览配置", padding="10")
+        preview_frame.pack(fill=tk.X, pady=5)
+        
+        self.create_entry(preview_frame, "preview_width", "预览宽度:", prefix="display")
+        self.create_entry(preview_frame, "preview_height", "预览高度:", prefix="display")
+        self.create_entry(preview_frame, "preview_columns", "预览列数:", prefix="display")
+        self.create_entry(preview_frame, "preview_rows", "预览行数:", prefix="display")
+
+        # 多窗口预览开关
+        multi_row = ttk.Frame(preview_frame)
+        multi_row.pack(fill=tk.X, pady=3)
+        ttk.Label(multi_row, text="多窗口预览:", width=15).pack(side=tk.LEFT)
+        self.preview_multi_window_var = tk.BooleanVar(value=False)
+        multi_cb = ttk.Checkbutton(multi_row, variable=self.preview_multi_window_var)
+        multi_cb.pack(side=tk.LEFT, padx=5)
+        self.config_vars["display_preview_multi_window"] = self.preview_multi_window_var
+
+        # 预览全部配置组开关
+        all_row = ttk.Frame(preview_frame)
+        all_row.pack(fill=tk.X, pady=3)
+        ttk.Label(all_row, text="预览全部配置组:", width=15).pack(side=tk.LEFT)
+        self.preview_use_all_groups_var = tk.BooleanVar(value=True)
+        all_cb = ttk.Checkbutton(all_row, variable=self.preview_use_all_groups_var)
+        all_cb.pack(side=tk.LEFT, padx=5)
+        self.config_vars["display_preview_use_all_groups"] = self.preview_use_all_groups_var
+        
+        # 保存按钮
+        save_frame = ttk.Frame(right_frame)
+        save_frame.pack(fill=tk.X, pady=10)
+        
+        ttk.Button(save_frame, text="保存配置", command=self.save_config).pack(side=tk.LEFT, padx=5)
+        ttk.Button(save_frame, text="启动预览", command=self.start_preview).pack(side=tk.LEFT, padx=5)
+        
+        # 启动程序按钮组
+        start_frame = ttk.Frame(right_frame)
+        start_frame.pack(fill=tk.X, pady=5)
+        
+        ttk.Button(start_frame, text="启动单个配置组", command=self.start_program).pack(side=tk.LEFT, padx=5)
+        ttk.Button(start_frame, text="启动多个配置组", command=self.start_multi_program).pack(side=tk.LEFT, padx=5)
+    
+    def create_entry(self, parent, key, label, prefix=None):
+        """创建输入框"""
+        frame = ttk.Frame(parent)
+        frame.pack(fill=tk.X, pady=3)
+        
+        ttk.Label(frame, text=label, width=15).pack(side=tk.LEFT)
+        
+        var = tk.StringVar()
+        entry = ttk.Entry(frame, textvariable=var, width=25)
+        entry.pack(side=tk.LEFT, fill=tk.X, expand=True, padx=5)
+        
+        # 保存变量引用
+        if prefix:
+            full_key = f"{prefix}_{key}"
+        else:
+            full_key = key
+        self.config_vars[full_key] = var
+    
+    def load_current_config(self):
+        """加载当前配置到界面"""
+        # 加载配置组列表
+        self.update_group_list()
+        
+        # 如果有配置组，默认选择第一个
+        if len(config_manager.config['groups']) > 0:
+            self.selected_index = 0
+            self.group_listbox.selection_set(0)
+            self.load_group_config(0)
+        
+        # 加载预览配置
+        display = config_manager.config.get('display', {})
+        for key in ['preview_width', 'preview_height', 'preview_columns', 'preview_rows']:
+            if key in display:
+                self.config_vars[f"display_{key}"].set(str(display[key]))
+        # 多窗口预览
+        if 'preview_multi_window' in display:
+            self.preview_multi_window_var.set(bool(display['preview_multi_window']))
+        # 预览全部配置组
+        if 'preview_use_all_groups' in display:
+            self.preview_use_all_groups_var.set(bool(display['preview_use_all_groups']))
+        # 初次自动扫描一次采集卡和串口
+        try:
+            self.scan_cameras()
+        except Exception:
+            pass
+        try:
+            self.scan_ports()
+        except Exception:
+            pass
+    
+    def update_group_list(self):
+        """更新配置组列表"""
+        self.group_listbox.delete(0, tk.END)
+        for i, group in enumerate(config_manager.config['groups']):
+            status = "✓" if group.get('active', False) else "  "
+            self.group_listbox.insert(tk.END, f"{status} {group['name']}")
+    
+    def on_group_select(self, event):
+        """选择配置组"""
+        selection = self.group_listbox.curselection()
+        if selection:
+            self.selected_index = selection[0]
+            self.load_group_config(self.selected_index)
+    
+    def load_group_config(self, index):
+        """加载配置组详情"""
+        group = config_manager.get_group_by_index(index)
+        if group:
+            for key, var in self.config_vars.items():
+                if not key.startswith('display_'):
+                    if key in group:
+                        var.set(str(group[key]))
+        # 确保下拉框列表与当前值一致
+        if group and 'camera_index' in group:
+            # 如果当前索引不在选项里，追加
+            values = list(self.camera_index_cb.cget('values')) if self.camera_index_cb.cget('values') else []
+            display_value = f"{group['camera_index']}"
+            if display_value not in values:
+                values.append(display_value)
+                self.camera_index_cb['values'] = values
+            self.camera_index_cb.set(display_value)
+        
+        # 确保串口下拉框列表与当前值一致
+        if group and 'serial_port' in group:
+            values = list(self.serial_port_cb.cget('values')) if self.serial_port_cb.cget('values') else []
+            port_value = group['serial_port']
+            
+            # 如果当前值不在列表中，尝试通过设备名匹配
+            matched = False
+            if hasattr(self, 'port_display_map'):
+                for display_text, device in self.port_display_map.items():
+                    if device == port_value:
+                        if display_text not in values:
+                            values.append(display_text)
+                        self.serial_port_cb.set(display_text)
+                        matched = True
+                        break
+            
+            if not matched:
+                # 如果找不到匹配，直接添加原始值
+                if port_value not in values:
+                    values.append(port_value)
+                    if not hasattr(self, 'port_display_map'):
+                        self.port_display_map = {}
+                    self.port_display_map[port_value] = port_value
+                self.serial_port_cb.set(port_value)
+
+    def scan_cameras(self, max_index: int = 10):
+        """扫描系统可用的采集卡索引，并填充下拉框"""
+        import warnings
+        import sys
+        import io
+        found = []
+        
+        # 临时设置OpenCV日志级别（兼容不同版本）
+        import os
+        old_level = os.environ.get('OPENCV_LOG_LEVEL', '')
+        os.environ['OPENCV_LOG_LEVEL'] = 'SILENT'  # 尝试更严格的级别
+        os.environ['OPENCV_IO_ENABLE_OPENEXR'] = '0'
+        
+        # 尝试设置日志级别（不同版本的OpenCV API不同）
+        try:
+            if hasattr(cv2, 'setLogLevel'):
+                # OpenCV 4.x
+                if hasattr(cv2, 'LOG_LEVEL_SILENT'):
+                    cv2.setLogLevel(cv2.LOG_LEVEL_SILENT)
+                elif hasattr(cv2, 'LOG_LEVEL_ERROR'):
+                    cv2.setLogLevel(cv2.LOG_LEVEL_ERROR)
+                elif hasattr(cv2, 'utils'):
+                    cv2.utils.setLogLevel(0)  # 0=SILENT/ERROR级别
+        except Exception:
+            pass
+        
+        # 重定向stderr来捕获OpenCV的错误输出
+        old_stderr = sys.stderr
+        suppressed_output = io.StringIO()
+        
+        try:
+            # 暂时重定向stderr以抑制OpenCV的错误消息
+            sys.stderr = suppressed_output
+            
+            for idx in range(max_index + 1):
+                cap = None
+                try:
+                    with warnings.catch_warnings():
+                        warnings.filterwarnings('ignore')
+                        # 尝试DSHOW后端
+                        cap = cv2.VideoCapture(idx, cv2.CAP_DSHOW)
+                        if not cap.isOpened():
+                            # 回退默认后端再试
+                            cap = cv2.VideoCapture(idx)
+                        
+                        if cap.isOpened():
+                            # 测试读取一帧，确保真正可用
+                            ret, test_frame = cap.read()
+                            if ret and test_frame is not None:
+                                found.append(str(idx))
+                            cap.release()
+                except Exception:
+                    if cap:
+                        try:
+                            cap.release()
+                        except:
+                            pass
+                    continue
+        finally:
+            # 恢复stderr
+            sys.stderr = old_stderr
+            # 恢复原来的日志级别
+            if old_level:
+                os.environ['OPENCV_LOG_LEVEL'] = old_level
+            else:
+                os.environ.pop('OPENCV_LOG_LEVEL', None)
+            os.environ.pop('OPENCV_IO_ENABLE_OPENEXR', None)
+        
+        if not found:
+            found = ["0"]  # 至少给一个默认项，避免为空
+            messagebox.showwarning("扫描完成", "未发现可用采集卡，已添加默认选项 0")
+        else:
+            messagebox.showinfo("扫描完成", f"发现可用采集卡索引: {', '.join(found)}")
+        
+        self.camera_index_cb['values'] = found
+        # 若当前无选择，则选择第一项
+        if not self.camera_index_var.get() and found:
+            self.camera_index_cb.set(found[0])
+    
+    def scan_ports(self):
+        """扫描系统可用的串口，并填充下拉框（显示设备描述）"""
+        from utils.device_scanner import scan_serial_ports_with_info
+        
+        found_real = []
+        port_display_map = {}  # 显示文本 -> 实际设备名
+        
+        try:
+            ports_info = scan_serial_ports_with_info()
+            for port_info in ports_info:
+                device = port_info['device']
+                description = port_info.get('description', '')
+                hwid = port_info.get('hwid', '')
+                
+                # 显示格式：COM3 (设备描述) [HWID]
+                if description:
+                    display_text = f"{device} ({description})"
+                else:
+                    display_text = device
+                
+                # 如果有HWID，添加到显示中（较短版本）
+                if hwid:
+                    # 提取VID/PID部分（如果存在）
+                    if 'VID_' in hwid and 'PID_' in hwid:
+                        import re
+                        vid_match = re.search(r'VID_([0-9A-F]+)', hwid)
+                        pid_match = re.search(r'PID_([0-9A-F]+)', hwid)
+                        if vid_match and pid_match:
+                            display_text += f" [{vid_match.group(1)}:{pid_match.group(1)}]"
+                
+                found_real.append(display_text)
+                port_display_map[display_text] = device
+        except Exception as e:
+            print(f"扫描串口错误: {e}")
+            # 回退到简单扫描
+            try:
+                ports = serial.tools.list_ports.comports()
+                found_real = [port.device for port in ports]
+                found_real.sort(key=lambda x: int(x.replace('COM', '')) if x.replace('COM', '').isdigit() else 999)
+                port_display_map = {p: p for p in found_real}
+            except:
+                pass
+        
+        # 如果没有发现实际端口，使用默认端口列表
+        if not found_real:
+            found = ["COM1", "COM2", "COM3", "COM4", "COM5", "COM6"]
+            port_display_map = {p: p for p in found}
+            messagebox.showwarning("警告", "未发现可用串口设备，已添加常用默认选项")
+        else:
+            found = found_real
+            messagebox.showinfo("扫描完成", f"发现 {len(found)} 个串口设备\n\n设备信息已包含描述和硬件ID")
+        
+        # 保存映射关系，用于后续获取实际设备名
+        self.port_display_map = port_display_map
+        
+        self.serial_port_cb['values'] = found
+        # 若当前无选择，则选择第一项
+        if not self.serial_port_var.get() and found:
+            self.serial_port_cb.set(found[0])
+    
+    def add_group(self):
+        """添加配置组"""
+        name = simpledialog.askstring("添加配置组", "请输入配置组名称:", initialvalue=f"配置组{len(config_manager.config['groups'])+1}")
+        if name:
+            config_manager.add_group(name)
+            self.update_group_list()
+            messagebox.showinfo("成功", f"已添加配置组: {name}")
+    
+    def delete_group(self):
+        """删除配置组"""
+        selection = self.group_listbox.curselection()
+        if not selection:
+            messagebox.showwarning("警告", "请先选择要删除的配置组")
+            return
+        
+        if messagebox.askyesno("确认", "确定要删除选中的配置组吗？"):
+            config_manager.delete_group(selection[0])
+            self.update_group_list()
+            messagebox.showinfo("成功", "配置组已删除")
+    
+    def set_active(self):
+        """设置为活动配置组"""
+        selection = self.group_listbox.curselection()
+        if not selection:
+            messagebox.showwarning("警告", "请先选择要设为活动的配置组")
+            return
+        
+        config_manager.set_active_group(selection[0])
+        self.update_group_list()
+        messagebox.showinfo("成功", "已设置为活动配置组")
+    
+    def save_config(self):
+        """保存当前编辑的配置"""
+        # 检查索引有效性
+        if self.selected_index < 0 or self.selected_index >= len(config_manager.config['groups']):
+            messagebox.showerror("错误", f"请先选择一个有效的配置组")
+            return False
+            
+        # 保存当前组的配置
+        group = config_manager.get_group_by_index(self.selected_index)
+        if not group:
+            messagebox.showerror("错误", f"配置组不存在")
+            return False
+            
+        for key, var in self.config_vars.items():
+            if not key.startswith('display_'):
+                try:
+                    value_str = var.get().strip() if var.get() else ""
+                    
+                    # 特殊处理：某些字段需要转换为整数
+                    if key in ['camera_index', 'camera_width', 'camera_height', 'serial_baudrate', 'move_velocity']:
+                        try:
+                            # 如果为空，使用当前值或默认值
+                            if value_str:
+                                value = int(value_str)
+                            else:
+                                # 如果下拉框为空，尝试从当前配置获取
+                                value = group.get(key, 0)
+                        except ValueError:
+                            # 转换失败，使用当前值
+                            value = group.get(key, 0)
+                            print(f"⚠️ 字段 {key} 的值 '{value_str}' 无效，使用当前值 {value}")
+                    else:
+                        # 字符串字段
+                        value = value_str if value_str else group.get(key, '')
+                        # 特殊处理串口：从显示文本中提取实际设备名
+                        if key == 'serial_port' and hasattr(self, 'port_display_map'):
+                            # 如果值是显示文本，转换为实际设备名
+                            if value in self.port_display_map:
+                                value = self.port_display_map[value]
+                    
+                    group[key] = value
+                except Exception as e:
+                    print(f"保存字段 {key} 时出错: {e}")
+                    import traceback
+                    traceback.print_exc()
+                    # 保存失败时使用当前值
+                    if key in group:
+                        pass  # 保持原值不变
+        
+        # 保存预览配置
+        display = config_manager.config.get('display', {})
+        for key in ['preview_width', 'preview_height', 'preview_columns', 'preview_rows']:
+            var = self.config_vars.get(f"display_{key}")
+            if var:
+                try:
+                    display[key] = int(var.get())
+                except:
+                    pass
+        # 保存多窗口预览布尔值
+        display['preview_multi_window'] = bool(self.preview_multi_window_var.get())
+        # 保存预览全部配置组布尔值
+        display['preview_use_all_groups'] = bool(self.preview_use_all_groups_var.get())
+        config_manager.config['display'] = display
+        
+        # 保存到文件
+        if config_manager.save_config():
+            # 更新左侧列表显示
+            self.update_group_list()
+            messagebox.showinfo("成功", "配置已保存")
+            return True
+        else:
+            messagebox.showerror("错误", "配置保存失败")
+            return False
+    
+    def start_preview(self):
+        """启动预览窗口（主线程，弹窗模式）"""
+        if not self.save_config_silent():
+            messagebox.showerror("错误", "配置保存失败，无法启动预览")
+            return
+        config_manager.load_config()
+        from preview import PreviewWindow
+        PreviewWindow().run(self.root)  # 以主窗口为父Toplevel弹出独立预览窗
+    
+    def save_config_silent(self):
+        """静默保存配置（不显示消息框）"""
+        # 保存当前组的配置
+        if self.selected_index < 0 or self.selected_index >= len(config_manager.config['groups']):
+            print(f"⚠️ 选中索引 {self.selected_index} 无效")
+            return False
+            
+        group = config_manager.get_group_by_index(self.selected_index)
+        if not group:
+            print(f"⚠️ 配置组不存在")
+            return False
+            
+        for key, var in self.config_vars.items():
+            if not key.startswith('display_'):
+                try:
+                    value_str = var.get().strip() if var.get() else ""
+                    
+                    # 特殊处理：某些字段需要转换为整数
+                    if key in ['camera_index', 'camera_width', 'camera_height', 'serial_baudrate', 'move_velocity']:
+                        try:
+                            # 如果为空，使用当前值或默认值
+                            if value_str:
+                                value = int(value_str)
+                            else:
+                                # 如果下拉框为空，尝试从当前配置获取
+                                value = group.get(key, 0)
+                        except ValueError:
+                            # 转换失败，使用当前值
+                            value = group.get(key, 0)
+                            print(f"⚠️ 字段 {key} 的值 '{value_str}' 无效，使用当前值 {value}")
+                    else:
+                        # 字符串字段
+                        value = value_str if value_str else group.get(key, '')
+                        # 特殊处理串口：从显示文本中提取实际设备名
+                        if key == 'serial_port' and hasattr(self, 'port_display_map'):
+                            # 如果值是显示文本，转换为实际设备名
+                            if value in self.port_display_map:
+                                value = self.port_display_map[value]
+                    
+                    group[key] = value
+                except Exception as e:
+                    print(f"保存字段 {key} 时出错: {e}")
+                    import traceback
+                    traceback.print_exc()
+                    # 保存失败时保持原值不变
+        
+        # 保存预览配置
+        display = config_manager.config.get('display', {})
+        for key in ['preview_width', 'preview_height', 'preview_columns', 'preview_rows']:
+            var = self.config_vars.get(f"display_{key}")
+            if var:
+                try:
+                    display[key] = int(var.get())
+                except:
+                    pass
+        # 保存多窗口预览布尔值（静默）
+        display['preview_multi_window'] = bool(self.preview_multi_window_var.get())
+        # 保存预览全部配置组布尔值（静默）
+        display['preview_use_all_groups'] = bool(self.preview_use_all_groups_var.get())
+        config_manager.config['display'] = display
+        
+        # 保存到文件
+        result = config_manager.save_config()
+        
+        # 更新左侧列表显示
+        if result:
+            self.update_group_list()
+        
+        return result
+    
+    def start_program(self):
+        """启动单个配置组的主程序并自动弹出采集卡预览窗口"""
+        # 保存配置（静默）
+        if not self.save_config_silent():
+            messagebox.showerror("错误", "配置保存失败")
+            return
+        
+        # 检查是否有活动配置组
+        active_groups = [g for g in config_manager.config['groups'] if g.get('active', False)]
+        if not active_groups:
+            messagebox.showwarning("警告", "没有活动的配置组\n\n请先选择一个配置组并设置为活动")
+            return
+        
+        # 启动单个配置组
+        import subprocess
+        import sys
+        try:
+            # 找到活动配置组的索引
+            active_group = active_groups[0]
+            group_index = config_manager.config['groups'].index(active_group)
+            
+            subprocess.Popen([
+                sys.executable, 
+                "main_single.py",
+                str(group_index)
+            ], creationflags=subprocess.CREATE_NEW_CONSOLE if sys.platform == 'win32' else 0)
+            
+            messagebox.showinfo("成功", f"已启动配置组: {active_group['name']}\n\n请在控制台查看运行状态")
+            # 成功后弹出预览
+            self.start_preview()
+        except Exception as e:
+            messagebox.showerror("错误", f"启动失败: {e}")
+    
+    def start_multi_program(self):
+        """启动多个配置组的主程序并自动弹出采集卡预览窗口"""
+        # 保存配置（静默）
+        if not self.save_config_silent():
+            messagebox.showerror("错误", "配置保存失败")
+            return
+        
+        # 启动多配置组管理器
+        import subprocess
+        import sys
+        try:
+            subprocess.Popen([
+                sys.executable, 
+                "main_multi.py"
+            ], creationflags=subprocess.CREATE_NEW_CONSOLE if sys.platform == 'win32' else 0)
+            
+            messagebox.showinfo("提示", "多配置组启动器已打开\n\n请在控制台中选择要启动的配置组")
+            # 成功后弹出预览
+            self.start_preview()
+        except Exception as e:
+            messagebox.showerror("错误", f"启动失败: {e}")
+    
+    def run(self):
+        """运行GUI"""
+        self.root.mainloop()
+
+if __name__ == "__main__":
+    app = ConfigGUI()
+    app.run()

launcher.py

@@ -0,0 +1,79 @@
+"""
+启动器 - 选择启动配置界面或主程序
+"""
+import tkinter as tk
+from tkinter import ttk
+import subprocess
+import sys
+
+def start_config_gui():
+    """启动配置界面"""
+    subprocess.Popen([sys.executable, "gui_config.py"])
+
+def start_preview():
+    """启动预览窗口"""
+    subprocess.Popen([sys.executable, "preview.py"])
+
+def start_main_program():
+    """启动主程序"""
+    subprocess.Popen([sys.executable, "main.py"])
+
+def create_launcher():
+    """创建启动器界面"""
+    root = tk.Tk()
+    root.title("火炬之光自动化 - 启动器")
+    root.geometry("400x580")
+    root.resizable(False, False)
+    
+    # 标题
+    title_frame = tk.Frame(root)
+    title_frame.pack(fill=tk.X, pady=20)
+    
+    tk.Label(title_frame, text="火炬之光自动化脚本", font=('Arial', 16, 'bold')).pack()
+    tk.Label(title_frame, text="请选择要启动的功能", font=('Arial', 10)).pack()
+    
+    # 按钮区域
+    button_frame = tk.Frame(root)
+    button_frame.pack(fill=tk.BOTH, expand=True, padx=40, pady=20)
+    
+    # 配置界面按钮
+    config_btn = tk.Button(button_frame, text="⚙️ 配置管理", 
+                          command=start_config_gui,
+                          width=30, height=3, font=('Arial', 12))
+    config_btn.pack(pady=10)
+    
+    # 预览窗口按钮
+    preview_btn = tk.Button(button_frame, text="👁️ 采集卡预览", 
+                           command=start_preview,
+                           width=30, height=3, font=('Arial', 12))
+    preview_btn.pack(pady=10)
+    
+    # 主程序按钮
+    main_btn = tk.Button(button_frame, text="🚀 启动单个配置组", 
+                         command=start_main_program,
+                         width=30, height=2, font=('Arial', 11))
+    main_btn.pack(pady=5)
+    
+    def start_multi_program():
+        """启动多配置组程序"""
+        subprocess.Popen([sys.executable, "main_multi.py"], 
+                        creationflags=subprocess.CREATE_NEW_CONSOLE if sys.platform == 'win32' else 0)
+    
+    # 多配置组启动按钮
+    multi_btn = tk.Button(button_frame, text="🔥 启动多个配置组", 
+                          command=start_multi_program,
+                          width=30, height=2, font=('Arial', 11))
+    multi_btn.pack(pady=5)
+    
+    # 说明
+    info_label = tk.Label(root, 
+                         text="提示：首次使用请先点击配置管理设置参数",
+                         font=('Arial', 9), 
+                         fg='gray')
+    info_label.pack(side=tk.BOTTOM, pady=10)
+    
+    root.mainloop()
+
+if __name__ == "__main__":
+    create_launcher()
+

main.py

@@ -1,20 +1,68 @@
 import cv2
-from utils.get_image import get_image
-from utils.mouse import mouse_gui
+from utils.get_image import GetImage
+from utils.mouse import init_mouse_keyboard, Mouse_guiji
 from ultralytics import YOLO
 import time
 import serial
 import ch9329Comm
-import time
 import random
 import math
 from utils import shizi
+from config import config_manager
+
+# 加载YOLO模型
 model = YOLO(r"best.pt").to('cuda')
 model0 = YOLO(r"best0.pt").to('cuda')
 
+# 从配置加载
+active_group = config_manager.get_active_group()
+if active_group is None:
+    print("❌ 没有活动的配置组，请在gui_config.py中设置")
+    exit(1)
+
+print(f"📋 使用配置组: {active_group['name']}")
+
+# 初始化串口和鼠标
+init_mouse_keyboard(active_group)
+
+# 初始化键盘和鼠标
 keyboard = ch9329Comm.keyboard.DataComm()
-mouse = ch9329Comm.mouse.DataComm(1920, 1080)
-kong_detections = {
+from utils.mouse import mouse, mouse_gui  # 导入已初始化的mouse和mouse_gui
+
+# 创建全局的mouse_gui实例
+mouse_gui = Mouse_guiji()
+
+# 初始化采集卡
+get_image = GetImage(
+    cam_index=active_group['camera_index'],
+    width=active_group['camera_width'],
+    height=active_group['camera_height']
+)
+
+# 检查采集卡是否初始化成功
+if get_image.cap is None:
+    print(f"❌ 采集卡 {active_group['camera_index']} 初始化失败")
+    print("请检查：")
+    print("1. 采集卡是否正确连接")
+    print("2. 采集卡索引是否正确")
+    print("3. 采集卡驱动是否安装")
+    exit(1)
+
+print(f"✅ 初始化完成 - 串口:{active_group['serial_port']} 采集卡:{active_group['camera_index']}")
+
+# 全局变量
+left = 0
+top = 30
+k = 0  # 控制转圈的方向
+panduan = False  # 是否在图内
+boss_pd = False  # 是否到boss关卡
+rw = (632, 378)
+
+# 从配置读取移动速度
+v = active_group['move_velocity']
+
+def yolo_shibie(im_PIL, detections, model):
+    detections = {
         'center': None,
         'next': None,
         'npc1': None,
@@ -26,39 +74,34 @@ kong_detections = {
         'gw': [],
         'zhaozi': None
     }
-left=0
-top=30
-k=0#控制转圈的方向
-panduan=False#是否在图内
-boss_pd=False#是否到boss关卡
-rw=(632,378)
-def yolo_shibie(im_PIL,detections,model):
-    results = model(im_PIL)#目标检测
+    results = model(im_PIL)  # 目标检测
     for result in results:
         for i in range(len(result.boxes.xyxy)):
             left, top, right, bottom = result.boxes.xyxy[i]
             scalar_tensor = result.boxes.cls[i]
             value = scalar_tensor.item()
             label = result.names[int(value)]
-            if label=='center'or label=='next' or label=='boss' or label=='zhaozi':
-                player_x = int(left+(right-left)/2)
-                player_y = int(top+(bottom-top)/2)+30
+            if label == 'center' or label == 'next' or label == 'boss' or label == 'zhaozi':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(top + (bottom - top) / 2) + 30
                 RW = [player_x, player_y]
                 detections[label] = RW
-            elif label=='daojv' or label=='gw':
+            elif label == 'daojv' or label == 'gw':
                 player_x = int(left + (right - left) / 2)
                 player_y = int(top + (bottom - top) / 2) + 30
                 RW = [player_x, player_y]
                 detections[label].append(RW)
-            elif label=='npc1' or label=='npc2' or label=='npc3' or label=='npc4':
-                player_x = int(left+(right-left)/2)
-                player_y = int(bottom)+30
+            elif label == 'npc1' or label == 'npc2' or label == 'npc3' or label == 'npc4':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(bottom) + 30
                 RW = [player_x, player_y]
                 detections[label] = RW
     return detections
+
 def sq(p1, p2):
     """计算两点之间的欧式距离"""
     return math.sqrt((p1[0] - p2[0]) ** 2 + (p1[1] - p2[1]) ** 2)
+
 def process_points(points):
     if not points:
         return None  # 空列表情况
@@ -89,9 +132,10 @@ def process_points(points):
                 best_point = p
 
         return best_point
+
 def move_randomly(rw, k):
     k = k % 4
-    suiji_t=float(random.randint(10,13)/10)
+    suiji_t = float(random.randint(10, 13) / 10)
     if k == 0:
         keyboard.send_data("66")
         time.sleep(suiji_t)
@@ -109,18 +153,21 @@ def move_randomly(rw, k):
         time.sleep(suiji_t)
         keyboard.release()  # Release
     return k + 1
-def move_to(rw,mb):
-    v=470
-    if rw[0]>=mb[0]:
+
+def move_to(rw, mb):
+    """使用配置的v值"""
+    global v
+    v = active_group['move_velocity']  # 每次都从配置读取最新值
+    if rw[0] >= mb[0]:
         keyboard.send_data("44")
-        time.sleep(float(abs(rw[0]-mb[0])/v))
+        time.sleep(float(abs(rw[0] - mb[0]) / v))
         keyboard.release()  # Release
     else:
         keyboard.send_data("66")
         time.sleep(float(abs(rw[0] - mb[0]) / v))
         keyboard.release()  # Release
 
-    if rw[1]>=mb[1]:
+    if rw[1] >= mb[1]:
         keyboard.send_data("88")
         time.sleep(float(abs(rw[1] - mb[1]) / v))
         keyboard.release()  # Release
@@ -128,7 +175,9 @@ def move_to(rw,mb):
         keyboard.send_data("22")
         time.sleep(float(abs(rw[1] - mb[1]) / v))
         keyboard.release()
-i=0
+
+i = 0
+print("🚀 开始自动化...")
 while True:
     detections = {
         'center': None,
@@ -140,13 +189,21 @@ while True:
         'boss': None,
         'daojv': [],
         'gw': [],
-        'zhaozi':None
-
+        'zhaozi': None
     }
-    im_opencv = get_image.get_frame()#[RGB,PIL]
-    detections=yolo_shibie(im_opencv[1],detections,model)
-    if   shizi.tuwai(im_opencv[0]):  # 进图算法
     im_opencv = get_image.get_frame()  # [RGB,PIL]
+    if im_opencv is None:
+        print("⚠️ 无法获取图像帧，跳过本次循环")
+        time.sleep(0.1)
+        continue
+    detections = yolo_shibie(im_opencv[1], detections, model)
+    
+    if shizi.tuwai(im_opencv[0]) or yolo_shibie(im_opencv[1], model0)['npc1'] is not None or yolo_shibie(im_opencv[1], model0)['npc2'] is not None or yolo_shibie(im_opencv[1], model0)['npc3'] is not None or yolo_shibie(im_opencv[1], model0)['npc4'] is not None:  # 进图算法  # 进图算法
+        im_opencv = get_image.get_frame()  # [RGB,PIL]
+        if im_opencv is None:
+            print("⚠️ 无法获取图像帧，跳过本次循环")
+            time.sleep(0.1)
+            continue
         detections = yolo_shibie(im_opencv[1], detections, model0)
         print('当前在城镇中')
         if detections['npc1'] is not None and sq(rw, detections['npc1']) > 80:
@@ -162,72 +219,189 @@ while True:
             continue
         elif detections['npc3'] is not None and detections['npc4'] is None:
             print("在npc3旁边，向右走")
-            mb = (rw[0], detections['npc3'][1]-50)
+            mb = (rw[0], detections['npc3'][1] - 50)
             move_to(rw, mb)
             mb = (rw[0] + 700, rw[1])
             move_to(rw, mb)
             continue
         elif detections['npc4'] is not None:
-            if sq(detections['npc4'], rw) < 50:
-                print("离npc4很近 直接进入")
+            npc4_pos = detections['npc4']
+            current_distance = sq(npc4_pos, rw)
+            
+            # 如果已经非常接近，直接进入
+            if current_distance < 30:
+                print(f"离npc4很近 (距离={current_distance:.1f})，直接进入")
                 keyboard.send_data("DD")
                 time.sleep(0.15)
                 keyboard.release()
                 time.sleep(1)
                 im_opencv = get_image.get_frame()  # [RGB,PIL]
+                if im_opencv is None:
+                    print("⚠️ 无法获取图像帧")
+                    continue
                 if shizi.daoying(im_opencv[0]):
                     mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
                     time.sleep(1)
                 continue
-            else:
-                print("离npc4有点远 点击进入")
             
-                move_to(rw, detections['npc4'])
+            # 循环靠近npc4，直到足够接近或达到最大尝试次数
+            print(f"开始靠近npc4，当前距离={current_distance:.1f}")
+            max_attempts = 15  # 最大尝试次数
+            attempt = 0
+            last_distance = current_distance
+            stuck_count = 0  # 卡住计数器
+            
+            while attempt < max_attempts:
+                attempt += 1
+                
+                # 重新获取当前位置和npc4位置
+                im_opencv = get_image.get_frame()
+                if im_opencv is None:
+                    print("⚠️ 无法获取图像帧")
+                    break
+                
+                detections_temp = {
+                    'center': None, 'next': None,
+                    'npc1': None, 'npc2': None, 'npc3': None, 'npc4': None,
+                    'boss': None, 'zhaozi': None,
+                    'daojv': [], 'gw': []
+                }
+                detections_temp = yolo_shibie(im_opencv[1], detections_temp, model0)
+                
+                # 如果npc4丢失，重新检测
+                if detections_temp['npc4'] is None:
+                    print(f"⚠️ npc4丢失，重新检测 (尝试 {attempt}/{max_attempts})")
+                    time.sleep(0.5)
+                    continue
+                
+                npc4_pos = detections_temp['npc4']
+                current_distance = sq(npc4_pos, rw)
+                
+                print(f"  尝试 {attempt}/{max_attempts}: 距离npc4={current_distance:.1f}")
+                
+                # 如果已经足够接近，退出循环
+                if current_distance < 35:
+                    print(f"✅ 已足够接近npc4 (距离={current_distance:.1f})")
+                    break
+                
+                # 检测是否卡住（距离不再减小）
+                if abs(current_distance - last_distance) < 5:
+                    stuck_count += 1
+                    if stuck_count >= 3:
+                        print(f"⚠️ 检测到卡住，尝试微调移动")
+                        # 尝试向npc4方向微调
+                        dx = npc4_pos[0] - rw[0]
+                        dy = npc4_pos[1] - rw[1]
+                        # 计算方向向量，但只移动一小段距离
+                        step_size = 50  # 小步移动
+                        distance_to_move = min(step_size, current_distance * 0.3)
+                        if abs(dx) > abs(dy):
+                            # 主要横向移动
+                            target_x = rw[0] + int(dx / abs(dx) * distance_to_move) if dx != 0 else rw[0]
+                            target_y = rw[1]
+                        else:
+                            # 主要纵向移动
+                            target_x = rw[0]
+                            target_y = rw[1] + int(dy / abs(dy) * distance_to_move) if dy != 0 else rw[1]
+                        mb = (target_x, target_y)
+                        move_to(rw, mb)
+                        time.sleep(0.3)
+                        stuck_count = 0
+                    else:
+                        # 正常移动，但使用小步长
+                        step_size = min(80, current_distance * 0.4)  # 移动距离为目标距离的40%，最多80像素
+                        dx = npc4_pos[0] - rw[0]
+                        dy = npc4_pos[1] - rw[1]
+                        # 归一化方向向量
+                        if abs(dx) > 0.1 or abs(dy) > 0.1:
+                            direction_mag = math.sqrt(dx*dx + dy*dy)
+                            target_x = rw[0] + int(dx / direction_mag * step_size)
+                            target_y = rw[1] + int(dy / direction_mag * step_size)
+                            mb = (target_x, target_y)
+                            move_to(rw, mb)
+                            time.sleep(0.2)  # 短暂等待，让角色移动
+                else:
+                    # 距离在减小，正常移动
+                    stuck_count = 0
+                    step_size = min(100, current_distance * 0.5)  # 移动距离为目标距离的50%，最多100像素
+                    dx = npc4_pos[0] - rw[0]
+                    dy = npc4_pos[1] - rw[1]
+                    # 归一化方向向量
+                    if abs(dx) > 0.1 or abs(dy) > 0.1:
+                        direction_mag = math.sqrt(dx*dx + dy*dy)
+                        target_x = rw[0] + int(dx / direction_mag * step_size)
+                        target_y = rw[1] + int(dy / direction_mag * step_size)
+                        mb = (target_x, target_y)
+                        move_to(rw, mb)
+                        time.sleep(0.2)  # 短暂等待，让角色移动
+                
+                last_distance = current_distance
+                
+                # 短暂延迟，让角色位置更新
+                time.sleep(0.3)
+            
+            # 移动完成后，检查是否可以进入
+            final_distance = sq(npc4_pos, rw)
+            print(f"靠近完成，最终距离={final_distance:.1f}")
+            
+            if final_distance < 50:
+                print("距离足够近，尝试进入")
+                keyboard.send_data("DD")
+                time.sleep(0.15)
+                keyboard.release()
                 time.sleep(1)
-                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                im_opencv = get_image.get_frame()
+                if im_opencv is None:
+                    print("⚠️ 无法获取图像帧")
+                    continue
+                if shizi.daoying(im_opencv[0]):
+                    mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                    time.sleep(1)
+            else:
+                print(f"⚠️ 未能足够接近npc4 (距离={final_distance:.1f})，尝试点击进入")
+                move_to(rw, npc4_pos)
+                time.sleep(1)
+                im_opencv = get_image.get_frame()
+                if im_opencv is None:
+                    print("⚠️ 无法获取图像帧")
+                    continue
                 if shizi.daoying(im_opencv[0]):
                     mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
                     time.sleep(1)
             continue
-    elif shizi.tiaozhan(im_opencv[0]):#开启挑战
+    elif shizi.tiaozhan(im_opencv[0]):  # 开启挑战
         print('进入塔4')
-        mouse_gui.send_data_absolute(left+1100,top+600,may=1)
+        mouse_gui.send_data_absolute(left + 1100, top + 600, may=1)
         time.sleep(0.3)
-        mouse_gui.send_data_absolute(left + 433, top + 455,may=1)
+        mouse_gui.send_data_absolute(left + 433, top + 455, may=1)
         panduan = True
         continue
-    elif shizi.jieshu(im_opencv[0]):#结束挑战
+    elif shizi.jieshu(im_opencv[0]):  # 结束挑战
         print('结束挑战')
-        mouse_gui.send_data_absolute(left+542,top+644,may=1)
+        mouse_gui.send_data_absolute(left + 542, top + 644, may=1)
         time.sleep(0.8)
-        mouse_gui.send_data_absolute(left + 706, top + 454,may=1)
+        mouse_gui.send_data_absolute(left + 706, top + 454, may=1)
         continue
-    elif panduan :#图内情况
+    elif panduan:  # 图内情况   
         print("在图内")
-        if shizi.shuzi(im_opencv[0]) :
+        if shizi.shuzi(im_opencv[0]):
             boss_pd = True
             print("进入到boss！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！")
         if shizi.fuhuo(im_opencv[0]):
             print('点击复活')
             mouse_gui.send_data_absolute(left + 536, top + 627, may=1)
-            mouse_gui.send_data_absolute(rw[0], rw[1], may=0)
-            continue
-        if detections['zhaozi'] is not None:
-            move_to(rw,detections['zhaozi'])
-            continue
-        if len(detections['daojv'])!=0:
-            move_to(rw,process_points(detections['daojv']))
-            for i in range(3+len(detections['daojv'])):
-                keyboard.send_data("AA")
             time.sleep(0.15)
             keyboard.release()
             continue
-        if shizi.tuichu(im_opencv[0])  and detections['next'] is None and len(detections['daojv'])==0  and len(detections['gw'])==0 and boss_pd:
+        if shizi.tuichu(im_opencv[0]) and detections['next'] is None and len(detections['daojv']) == 0 and len(detections['gw']) == 0 and boss_pd:
             print("识别到可以退出挑战！！！！！！！！！！！！！！！！！！")
             for i in range(3):
                 time.sleep(0.5)
 
-                im_opencv = get_image.get_frame()#[RGB,PIL]
+                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                if im_opencv is None:
+                    print("⚠️ 无法获取图像帧")
+                    continue
                 detections = {
                     'center': None,
                     'next': None,
@@ -239,52 +413,49 @@ while True:
                     'daojv': [],
                     'gw': [],
                     'zhaozi': None
-
                 }
-                detections = yolo_shibie(im_opencv[1], detections,model)
-                if detections['next'] is not None or len(detections['daojv'])!=0 or len(detections['gw'])!=0 or detections['boss'] is not None:
+                detections = yolo_shibie(im_opencv[1], detections, model)
+                if detections['next'] is not None or len(detections['daojv']) != 0 or len(detections['gw']) != 0 or detections['boss'] is not None:
                     break
             else:
-                mouse_gui.send_data_absolute(left + 640, top + 40,may=1)#点击退出
-                panduan = False#退出挑战
+                mouse_gui.send_data_absolute(left + 640, top + 40, may=1)  # 点击退出
+                panduan = False  # 退出挑战
                 boss_pd = False
                 time.sleep(2.0)
                 continue
-        if detections['center'] is None and detections['next'] is None and (len(detections['gw'])!=0 or detections["boss"] is not None):#识别不到中心情况 但是有怪物
+        if detections['center'] is None and detections['next'] is None and (len(detections['gw']) != 0 or detections["boss"] is not None):  # 识别不到中心情况 但是有怪物
             print("未检测到中心点，但是有怪物")
-            if len(detections['gw'])!=0:
-                move_to(rw,detections['gw'][0])
+            if len(detections['gw']) != 0:
+                move_to(rw, detections['gw'][0])
                 time.sleep(0.26)
-            elif detections['boss'] is not None:#跟随boss
-                boss_suiji1=random.randint(-30,30)
+            elif detections['boss'] is not None:  # 跟随boss
+                boss_suiji1 = random.randint(-30, 30)
                 boss_suiji2 = random.randint(-30, 30)
-                detections['boss']=(detections['boss'][0]+boss_suiji1,detections['boss'][1]+boss_suiji2)
-                move_to(rw,detections['boss'])
+                detections['boss'] = (detections['boss'][0] + boss_suiji1, detections['boss'][1] + boss_suiji2)
+                move_to(rw, detections['boss'])
                 time.sleep(0.7)
             continue
-        elif (detections['center'] is not None and detections['next'] is None and len(detections['gw'])!=0) or (boss_pd==True and detections['center'] is not None and detections['next'] is None)    :#识别到中心  但是有怪物
-            if detections['center'][0]>=rw[0] and detections['center'][1]<rw[1]:#3
-                mb=(rw[0]+100,rw[1])
-            elif detections['center'][0]<=rw[0] and detections['center'][1]<rw[1]:#4
-                mb=(rw[0], rw[1]-100)
-            elif detections['center'][0]<=rw[0] and detections['center'][1]>rw[1]:#1
-                mb=(rw[0]-100, rw[1])
-            elif detections['center'][0]>=rw[0] and detections['center'][1]>rw[1]:#2
-                mb=(rw[0], rw[1]+100)
+        elif (detections['center'] is not None and detections['next'] is None and len(detections['gw']) != 0) or (boss_pd == True and detections['center'] is not None and detections['next'] is None):  # 识别到中心  但是有怪物
+            if detections['center'][0] >= rw[0] and detections['center'][1] < rw[1]:  # 3
+                mb = (rw[0] + 100, rw[1])
+            elif detections['center'][0] <= rw[0] and detections['center'][1] < rw[1]:  # 4
+                mb = (rw[0], rw[1] - 100)
+            elif detections['center'][0] <= rw[0] and detections['center'][1] > rw[1]:  # 1
+                mb = (rw[0] - 100, rw[1])
+            elif detections['center'][0] >= rw[0] and detections['center'][1] > rw[1]:  # 2
+                mb = (rw[0], rw[1] + 100)
             move_to(rw, mb)
             time.sleep(0.25)
-
-
             continue
 
-        elif boss_pd==True and detections['center'] is None and detections['next'] is None:#boss出现了  但是没有中心
-            k=move_randomly(rw,k)
+        elif boss_pd == True and detections['center'] is None and detections['next'] is None:  # boss出现了  但是没有中心
+            k = move_randomly(rw, k)
             continue
 
         elif detections['next'] is not None:
             print('进入下一层啦啦啦啦啦啦')
             panduan = True
-            move_to(rw,detections['next'])
+            move_to(rw, detections['next'])
             for i in range(2):
                 keyboard.send_data("DD")
                 time.sleep(0.15)
@@ -298,3 +469,4 @@ while True:
         mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
         time.sleep(1)
         continue
+

main_multi.py

@@ -0,0 +1,163 @@
+"""
+多配置组启动器
+支持同时启动多个配置组的自动化程序
+"""
+import multiprocessing
+import sys
+from config import config_manager
+from main_single import run_automation_for_group
+from utils.logger import logger, throttle
+import logging
+import time
+
+
+def main():
+    """主函数"""
+    # 重新加载配置
+    config_manager.load_config()
+    
+    # 获取所有配置组
+    groups = config_manager.config.get('groups', [])
+    
+    if not groups:
+        logger.error("❌ 没有找到任何配置组")
+        logger.info("请先运行 gui_config.py 创建配置组")
+        return
+    
+    # 询问要启动哪些配置组
+    logger.info("=" * 60)
+    logger.info("🔥 多配置组启动器")
+    logger.info("=" * 60)
+    logger.info("\n可用配置组:")
+    for i, group in enumerate(groups):
+        active_mark = "✓" if group.get('active', False) else " "
+        logger.info(f"  [{i}] {active_mark} {group['name']}")
+        logger.info(f"      串口: {group['serial_port']} | 采集卡: {group['camera_index']}")
+    
+    logger.info("\n选择启动方式:")
+    logger.info("  1. 启动所有活动配置组")
+    logger.info("  2. 启动所有配置组")
+    logger.info("  3. 选择特定配置组")
+    logger.info("  0. 退出")
+    
+    choice = input("\n请选择 (0-3): ").strip()
+    
+    selected_indices = []
+    
+    if choice == "0":
+        logger.info("👋 退出")
+        return
+    elif choice == "1":
+        # 启动所有活动配置组
+        selected_indices = [i for i, g in enumerate(groups) if g.get('active', False)]
+        if not selected_indices:
+            logger.error("❌ 没有活动的配置组")
+            return
+        logger.info(f"\n✅ 将启动 {len(selected_indices)} 个活动配置组")
+    elif choice == "2":
+        # 启动所有配置组
+        selected_indices = list(range(len(groups)))
+        logger.info(f"\n✅ 将启动所有 {len(selected_indices)} 个配置组")
+    elif choice == "3":
+        # 选择特定配置组
+        indices_input = input("请输入要启动的配置组索引(用逗号分隔,如: 0,1,2): ").strip()
+        try:
+            selected_indices = [int(x.strip()) for x in indices_input.split(',')]
+            # 验证索引有效性
+            selected_indices = [i for i in selected_indices if 0 <= i < len(groups)]
+            if not selected_indices:
+                logger.error("❌ 没有有效的配置组索引")
+                return
+            logger.info(f"\n✅ 将启动 {len(selected_indices)} 个配置组")
+        except ValueError:
+            logger.error("❌ 输入格式错误")
+            return
+    else:
+        logger.error("❌ 无效选择")
+        return
+    
+    # 显示将要启动的配置组
+    logger.info("\n将要启动的配置组:")
+    for idx in selected_indices:
+        group = groups[idx]
+        logger.info(f"  • {group['name']} (串口:{group['serial_port']}, 采集卡:{group['camera_index']})")
+
+    # 串口冲突预检：同一串口被多个组占用通常会导致仅一路成功
+    port_to_groups = {}
+    for idx in selected_indices:
+        g = groups[idx]
+        port_to_groups.setdefault(g['serial_port'], []).append(g['name'])
+    conflicts = {p: names for p, names in port_to_groups.items() if p and len(names) > 1}
+    if conflicts:
+        logger.warning("⚠️ 检测到串口冲突（同一COM被多个组使用）：")
+        for p, names in conflicts.items():
+            logger.warning(f"  {p}: {', '.join(names)}")
+        go_on = input("上述冲突很可能导致仅一组成功，其它失败。仍要继续? (y/n): ").strip().lower()
+        if go_on != 'y':
+            logger.info("已取消启动以避免串口冲突")
+            return
+    
+    confirm = input("\n确认启动? (y/n): ").strip().lower()
+    if confirm != 'y':
+        logger.info("❌ 取消启动")
+        return
+    
+    # 启动多进程
+    processes = []
+    
+    for idx in selected_indices:
+        group = groups[idx]
+        logger.info(f"启动进程: {group['name']}...")
+        process = multiprocessing.Process(
+            target=run_automation_for_group,
+            args=(idx,),
+            name=f"Group-{idx}-{group['name']}"
+        )
+        process.start()
+        processes.append((idx, group['name'], process))
+        logger.info(f"✅ {group['name']} 已启动 (PID: {process.pid})")
+    
+    logger.info(f"\n✅ 成功启动 {len(processes)} 个配置组进程")
+    logger.info("\n" + "=" * 60)
+    logger.info("运行状态:")
+    logger.info("=" * 60)
+    
+    # 监控进程状态
+    try:
+        while True:
+            alive_count = 0
+            for idx, name, proc in processes:
+                if proc.is_alive():
+                    alive_count += 1
+                else:
+                    logger.warning(f"⚠️ {name} 进程已退出 (退出码: {proc.exitcode})")
+            
+            if alive_count == 0:
+                logger.info("\n所有进程已退出")
+                break
+            
+            time.sleep(2)
+            
+            # 打印存活状态
+            alive_names = [name for idx, name, proc in processes if proc.is_alive()]
+            if alive_names:
+                # 每2秒节流一次状态行
+                throttle("multi_alive", 2.0, logging.INFO, f"📊 运行中: {', '.join(alive_names)} ({alive_count}/{len(processes)})")
+                
+    except KeyboardInterrupt:
+        logger.warning("\n\n🛑 收到停止信号，正在关闭所有进程...")
+        for idx, name, proc in processes:
+            if proc.is_alive():
+                logger.info(f"正在停止 {name}...")
+                proc.terminate()
+                proc.join(timeout=5)
+                if proc.is_alive():
+                    logger.warning(f"强制停止 {name}...")
+                    proc.kill()
+                logger.info(f"✅ {name} 已停止")
+        logger.info("\n👋 所有进程已停止")
+
+if __name__ == "__main__":
+    multiprocessing.freeze_support()  # Windows下需要
+    main()
+

main_new.py

@@ -0,0 +1,325 @@
+import cv2
+from utils.get_image import GetImage
+from utils.mouse import init_mouse_keyboard, Mouse_guiji
+from ultralytics import YOLO
+import time
+import serial
+import ch9329Comm
+import random
+import math
+from utils import shizi
+from config import config_manager
+
+# 加载YOLO模型
+model = YOLO(r"best.pt").to('cuda')
+model0 = YOLO(r"best0.pt").to('cuda')
+
+# 从配置加载
+active_group = config_manager.get_active_group()
+if active_group is None:
+    print("❌ 没有活动的配置组，请在gui_config.py中设置")
+    exit(1)
+
+print(f"📋 使用配置组: {active_group['name']}")
+
+# 初始化串口和鼠标
+init_mouse_keyboard(active_group)
+
+# 初始化键盘和鼠标
+keyboard = ch9329Comm.keyboard.DataComm()
+from utils.mouse import mouse, mouse_gui  # 导入已初始化的mouse和mouse_gui
+
+# 创建全局的mouse_gui实例
+mouse_gui = Mouse_guiji()
+
+# 初始化采集卡
+get_image = GetImage(
+    cam_index=active_group['camera_index'],
+    width=active_group['camera_width'],
+    height=active_group['camera_height']
+)
+
+print(f"✅ 初始化完成 - 串口:{active_group['serial_port']} 采集卡:{active_group['camera_index']}")
+
+# 全局变量
+left = 0
+top = 30
+k = 0  # 控制转圈的方向
+panduan = False  # 是否在图内
+boss_pd = False  # 是否到boss关卡
+rw = (632, 378)
+
+# 从配置读取移动速度
+v = active_group['move_velocity']
+
+def yolo_shibie(im_PIL, detections, model):
+    results = model(im_PIL)  # 目标检测
+    for result in results:
+        for i in range(len(result.boxes.xyxy)):
+            left, top, right, bottom = result.boxes.xyxy[i]
+            scalar_tensor = result.boxes.cls[i]
+            value = scalar_tensor.item()
+            label = result.names[int(value)]
+            if label == 'center' or label == 'next' or label == 'boss' or label == 'zhaozi':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(top + (bottom - top) / 2) + 30
+                RW = [player_x, player_y]
+                detections[label] = RW
+            elif label == 'daojv' or label == 'gw':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(top + (bottom - top) / 2) + 30
+                RW = [player_x, player_y]
+                detections[label].append(RW)
+            elif label == 'npc1' or label == 'npc2' or label == 'npc3' or label == 'npc4':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(bottom) + 30
+                RW = [player_x, player_y]
+                detections[label] = RW
+    return detections
+
+def sq(p1, p2):
+    """计算两点之间的欧式距离"""
+    return math.sqrt((p1[0] - p2[0]) ** 2 + (p1[1] - p2[1]) ** 2)
+
+def process_points(points):
+    if not points:
+        return None  # 空列表情况
+
+    n = len(points)
+
+    if n == 1:
+        # 只有一个点，直接返回
+        return points[0]
+
+    elif n == 2:
+        # 两个点取中点
+        x = (points[0][0] + points[1][0]) / 2
+        y = (points[0][1] + points[1][1]) / 2
+        return [x, y]
+
+    else:
+        # 随机选3个点
+        sample_points = random.sample(points, 3)
+
+        # 对每个点计算到这3个点的总距离
+        min_sum = float('inf')
+        best_point = None
+        for p in points:
+            dist_sum = sum(sq(p, sp) for sp in sample_points)
+            if dist_sum < min_sum:
+                min_sum = dist_sum
+                best_point = p
+
+        return best_point
+
+def move_randomly(rw, k):
+    k = k % 4
+    suiji_t = float(random.randint(10, 13) / 10)
+    if k == 0:
+        keyboard.send_data("66")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    elif k == 1:
+        keyboard.send_data("88")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    elif k == 2:
+        keyboard.send_data("44")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    elif k == 3:
+        keyboard.send_data("22")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    return k + 1
+
+def move_to(rw, mb):
+    """使用配置的v值"""
+    global v
+    v = active_group['move_velocity']  # 每次都从配置读取最新值
+    if rw[0] >= mb[0]:
+        keyboard.send_data("44")
+        time.sleep(float(abs(rw[0] - mb[0]) / v))
+        keyboard.release()  # Release
+    else:
+        keyboard.send_data("66")
+        time.sleep(float(abs(rw[0] - mb[0]) / v))
+        keyboard.release()  # Release
+
+    if rw[1] >= mb[1]:
+        keyboard.send_data("88")
+        time.sleep(float(abs(rw[1] - mb[1]) / v))
+        keyboard.release()  # Release
+    else:
+        keyboard.send_data("22")
+        time.sleep(float(abs(rw[1] - mb[1]) / v))
+        keyboard.release()
+
+i = 0
+print("🚀 开始自动化...")
+while True:
+    detections = {
+        'center': None,
+        'next': None,
+        'npc1': None,
+        'npc2': None,
+        'npc3': None,
+        'npc4': None,
+        'boss': None,
+        'daojv': [],
+        'gw': [],
+        'zhaozi': None
+    }
+    im_opencv = get_image.get_frame()  # [RGB,PIL]
+    detections = yolo_shibie(im_opencv[1], detections, model)
+    
+    if shizi.tuwai(im_opencv[0]):  # 进图算法
+        im_opencv = get_image.get_frame()  # [RGB,PIL]
+        detections = yolo_shibie(im_opencv[1], detections, model0)
+        print('当前在城镇中')
+        if detections['npc1'] is not None and sq(rw, detections['npc1']) > 80:
+            print("向npc1靠近")
+            print(sq(rw, detections['npc1']))
+            move_to(rw, detections['npc1'])
+            continue
+        elif detections['npc1'] is not None and sq(rw, detections['npc1']) <= 80:
+            print("在npc旁边，向上走")
+            print(sq(rw, detections['npc1']))
+            mb = (detections['npc1'][0], detections['npc1'][1] - 1010)
+            move_to(rw, mb)
+            continue
+        elif detections['npc3'] is not None and detections['npc4'] is None:
+            print("在npc3旁边，向右走")
+            mb = (rw[0], detections['npc3'][1] - 50)
+            move_to(rw, mb)
+            mb = (rw[0] + 700, rw[1])
+            move_to(rw, mb)
+            continue
+        elif detections['npc4'] is not None:
+            if sq(detections['npc4'], rw) < 50:
+                print("离npc4很近 直接进入")
+                keyboard.send_data("DD")
+                time.sleep(0.15)
+                keyboard.release()
+                time.sleep(1)
+                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                if shizi.daoying(im_opencv[0]):
+                    mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                    time.sleep(1)
+                continue
+            else:
+                print("离npc4有点远 点击进入")
+                move_to(rw, detections['npc4'])
+                time.sleep(1)
+                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                if shizi.daoying(im_opencv[0]):
+                    mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                    time.sleep(1)
+                continue
+    elif shizi.tiaozhan(im_opencv[0]):  # 开启挑战
+        print('进入塔4')
+        mouse_gui.send_data_absolute(left + 1100, top + 600, may=1)
+        time.sleep(0.3)
+        mouse_gui.send_data_absolute(left + 433, top + 455, may=1)
+        panduan = True
+        continue
+    elif shizi.jieshu(im_opencv[0]):  # 结束挑战
+        print('结束挑战')
+        mouse_gui.send_data_absolute(left + 542, top + 644, may=1)
+        time.sleep(0.8)
+        mouse_gui.send_data_absolute(left + 706, top + 454, may=1)
+        continue
+    elif panduan:  # 图内情况
+        print("在图内")
+        if shizi.shuzi(im_opencv[0]):
+            boss_pd = True
+            print("进入到boss！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！")
+        if shizi.fuhuo(im_opencv[0]):
+            print('点击复活')
+            mouse_gui.send_data_absolute(left + 536, top + 627, may=1)
+            mouse_gui.send_data_absolute(rw[0], rw[1], may=0)
+            continue
+        if detections['zhaozi'] is not None:
+            move_to(rw, detections['zhaozi'])
+            continue
+        if len(detections['daojv']) != 0:
+            move_to(rw, process_points(detections['daojv']))
+            for i in range(3 + len(detections['daojv'])):
+                keyboard.send_data("AA")
+                time.sleep(0.15)
+                keyboard.release()
+            continue
+        if shizi.tuichu(im_opencv[0]) and detections['next'] is None and len(detections['daojv']) == 0 and len(detections['gw']) == 0 and boss_pd:
+            print("识别到可以退出挑战！！！！！！！！！！！！！！！！！！")
+            for i in range(3):
+                time.sleep(0.5)
+
+                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                detections = {
+                    'center': None,
+                    'next': None,
+                    'npc1': None,
+                    'npc2': None,
+                    'npc3': None,
+                    'npc4': None,
+                    'boss': None,
+                    'daojv': [],
+                    'gw': [],
+                    'zhaozi': None
+                }
+                detections = yolo_shibie(im_opencv[1], detections, model)
+                if detections['next'] is not None or len(detections['daojv']) != 0 or len(detections['gw']) != 0 or detections['boss'] is not None:
+                    break
+            else:
+                mouse_gui.send_data_absolute(left + 640, top + 40, may=1)  # 点击退出
+                panduan = False  # 退出挑战
+                boss_pd = False
+                time.sleep(2.0)
+                continue
+        if detections['center'] is None and detections['next'] is None and (len(detections['gw']) != 0 or detections["boss"] is not None):  # 识别不到中心情况 但是有怪物
+            print("未检测到中心点，但是有怪物")
+            if len(detections['gw']) != 0:
+                move_to(rw, detections['gw'][0])
+                time.sleep(0.26)
+            elif detections['boss'] is not None:  # 跟随boss
+                boss_suiji1 = random.randint(-30, 30)
+                boss_suiji2 = random.randint(-30, 30)
+                detections['boss'] = (detections['boss'][0] + boss_suiji1, detections['boss'][1] + boss_suiji2)
+                move_to(rw, detections['boss'])
+                time.sleep(0.7)
+            continue
+        elif (detections['center'] is not None and detections['next'] is None and len(detections['gw']) != 0) or (boss_pd == True and detections['center'] is not None and detections['next'] is None):  # 识别到中心  但是有怪物
+            if detections['center'][0] >= rw[0] and detections['center'][1] < rw[1]:  # 3
+                mb = (rw[0] + 100, rw[1])
+            elif detections['center'][0] <= rw[0] and detections['center'][1] < rw[1]:  # 4
+                mb = (rw[0], rw[1] - 100)
+            elif detections['center'][0] <= rw[0] and detections['center'][1] > rw[1]:  # 1
+                mb = (rw[0] - 100, rw[1])
+            elif detections['center'][0] >= rw[0] and detections['center'][1] > rw[1]:  # 2
+                mb = (rw[0], rw[1] + 100)
+            move_to(rw, mb)
+            time.sleep(0.25)
+            continue
+
+        elif boss_pd == True and detections['center'] is None and detections['next'] is None:  # boss出现了  但是没有中心
+            k = move_randomly(rw, k)
+            continue
+
+        elif detections['next'] is not None:
+            print('进入下一层啦啦啦啦啦啦')
+            panduan = True
+            move_to(rw, detections['next'])
+            for i in range(2):
+                keyboard.send_data("DD")
+                time.sleep(0.15)
+                keyboard.release()
+            continue
+
+        else:
+            k = move_randomly(rw, k)
+            continue
+    elif shizi.daoying(im_opencv[0]):
+        mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+        time.sleep(1)
+        continue
+

main_old.py

@@ -0,0 +1,300 @@
+import cv2
+from utils.get_image import get_image
+from utils.mouse import mouse_gui
+from ultralytics import YOLO
+import time
+import serial
+import ch9329Comm
+import time
+import random
+import math
+from utils import shizi
+model = YOLO(r"best.pt").to('cuda')
+model0 = YOLO(r"best0.pt").to('cuda')
+
+keyboard = ch9329Comm.keyboard.DataComm()
+mouse = ch9329Comm.mouse.DataComm(1920, 1080)
+kong_detections = {
+            'center': None,
+            'next': None,
+            'npc1': None,
+            'npc2': None,
+            'npc3': None,
+            'npc4': None,
+            'boss': None,
+            'daojv': [],
+            'gw': [],
+            'zhaozi': None
+        }
+left=0
+top=30
+k=0#控制转圈的方向
+panduan=False#是否在图内
+boss_pd=False#是否到boss关卡
+rw=(632,378)
+def yolo_shibie(im_PIL,detections,model):
+    results = model(im_PIL)#目标检测
+    for result in results:
+        for i in range(len(result.boxes.xyxy)):
+            left, top, right, bottom = result.boxes.xyxy[i]
+            scalar_tensor = result.boxes.cls[i]
+            value = scalar_tensor.item()
+            label = result.names[int(value)]
+            if label=='center'or label=='next' or label=='boss' or label=='zhaozi':
+                player_x = int(left+(right-left)/2)
+                player_y = int(top+(bottom-top)/2)+30
+                RW = [player_x, player_y]
+                detections[label] = RW
+            elif label=='daojv' or label=='gw':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(top + (bottom - top) / 2) + 30
+                RW = [player_x, player_y]
+                detections[label].append(RW)
+            elif label=='npc1' or label=='npc2' or label=='npc3' or label=='npc4':
+                player_x = int(left+(right-left)/2)
+                player_y = int(bottom)+30
+                RW = [player_x, player_y]
+                detections[label] = RW
+    return detections
+def sq(p1, p2):
+    """计算两点之间的欧式距离"""
+    return math.sqrt((p1[0] - p2[0]) ** 2 + (p1[1] - p2[1]) ** 2)
+def process_points(points):
+    if not points:
+        return None  # 空列表情况
+
+    n = len(points)
+
+    if n == 1:
+        # 只有一个点，直接返回
+        return points[0]
+
+    elif n == 2:
+        # 两个点取中点
+        x = (points[0][0] + points[1][0]) / 2
+        y = (points[0][1] + points[1][1]) / 2
+        return [x, y]
+
+    else:
+        # 随机选3个点
+        sample_points = random.sample(points, 3)
+
+        # 对每个点计算到这3个点的总距离
+        min_sum = float('inf')
+        best_point = None
+        for p in points:
+            dist_sum = sum(sq(p, sp) for sp in sample_points)
+            if dist_sum < min_sum:
+                min_sum = dist_sum
+                best_point = p
+
+        return best_point
+def move_randomly(rw, k):
+    k = k % 4
+    suiji_t=float(random.randint(10,13)/10)
+    if k == 0:
+        keyboard.send_data("66")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    elif k == 1:
+        keyboard.send_data("88")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    elif k == 2:
+        keyboard.send_data("44")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    elif k == 3:
+        keyboard.send_data("22")
+        time.sleep(suiji_t)
+        keyboard.release()  # Release
+    return k + 1
+def move_to(rw,mb):
+    v=470
+    if rw[0]>=mb[0]:
+        keyboard.send_data("44")
+        time.sleep(float(abs(rw[0]-mb[0])/v))
+        keyboard.release()  # Release
+    else:
+        keyboard.send_data("66")
+        time.sleep(float(abs(rw[0] - mb[0]) / v))
+        keyboard.release()  # Release
+
+    if rw[1]>=mb[1]:
+        keyboard.send_data("88")
+        time.sleep(float(abs(rw[1] - mb[1]) / v))
+        keyboard.release()  # Release
+    else:
+        keyboard.send_data("22")
+        time.sleep(float(abs(rw[1] - mb[1]) / v))
+        keyboard.release()
+i=0
+while True:
+    detections = {
+        'center': None,
+        'next': None,
+        'npc1': None,
+        'npc2': None,
+        'npc3': None,
+        'npc4': None,
+        'boss': None,
+        'daojv': [],
+        'gw': [],
+        'zhaozi':None
+
+    }
+    im_opencv = get_image.get_frame()#[RGB,PIL]
+    detections=yolo_shibie(im_opencv[1],detections,model)
+    if   shizi.tuwai(im_opencv[0]):  # 进图算法
+        im_opencv = get_image.get_frame()  # [RGB,PIL]
+        detections = yolo_shibie(im_opencv[1], detections, model0)
+        print('当前在城镇中')
+        if detections['npc1'] is not None and sq(rw, detections['npc1']) > 80:
+            print("向npc1靠近")
+            print(sq(rw, detections['npc1']))
+            move_to(rw, detections['npc1'])
+            continue
+        elif detections['npc1'] is not None and sq(rw, detections['npc1']) <= 80:
+            print("在npc旁边，向上走")
+            print(sq(rw, detections['npc1']))
+            mb = (detections['npc1'][0], detections['npc1'][1] - 1010)
+            move_to(rw, mb)
+            continue
+        elif detections['npc3'] is not None and detections['npc4'] is None:
+            print("在npc3旁边，向右走")
+            mb = (rw[0], detections['npc3'][1]-50)
+            move_to(rw, mb)
+            mb = (rw[0] + 700, rw[1])
+            move_to(rw, mb)
+            continue
+        elif detections['npc4'] is not None:
+            if sq(detections['npc4'], rw) < 50:
+                print("离npc4很近 直接进入")
+                keyboard.send_data("DD")
+                time.sleep(0.15)
+                keyboard.release()
+                time.sleep(1)
+                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                if shizi.daoying(im_opencv[0]):
+                    mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                    time.sleep(1)
+                continue
+            else:
+                print("离npc4有点远 点击进入")
+
+                move_to(rw, detections['npc4'])
+                time.sleep(1)
+                im_opencv = get_image.get_frame()  # [RGB,PIL]
+                if shizi.daoying(im_opencv[0]):
+                    mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                    time.sleep(1)
+                continue
+    elif shizi.tiaozhan(im_opencv[0]):#开启挑战
+        print('进入塔4')
+        mouse_gui.send_data_absolute(left+1100,top+600,may=1)
+        time.sleep(0.3)
+        mouse_gui.send_data_absolute(left + 433, top + 455,may=1)
+        panduan = True
+        continue
+    elif shizi.jieshu(im_opencv[0]):#结束挑战
+        print('结束挑战')
+        mouse_gui.send_data_absolute(left+542,top+644,may=1)
+        time.sleep(0.8)
+        mouse_gui.send_data_absolute(left + 706, top + 454,may=1)
+        continue
+    elif panduan :#图内情况
+        print("在图内")
+        if shizi.shuzi(im_opencv[0]) :
+            boss_pd = True
+            print("进入到boss！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！！")
+        if shizi.fuhuo(im_opencv[0]):
+            print('点击复活')
+            mouse_gui.send_data_absolute(left + 536, top + 627, may=1)
+            mouse_gui.send_data_absolute(rw[0], rw[1], may=0)
+            continue
+        if detections['zhaozi'] is not None:
+            move_to(rw,detections['zhaozi'])
+            continue
+        if len(detections['daojv'])!=0:
+            move_to(rw,process_points(detections['daojv']))
+            for i in range(3+len(detections['daojv'])):
+                keyboard.send_data("AA")
+                time.sleep(0.15)
+                keyboard.release()
+            continue
+        if shizi.tuichu(im_opencv[0])  and detections['next'] is None and len(detections['daojv'])==0  and len(detections['gw'])==0 and boss_pd:
+            print("识别到可以退出挑战！！！！！！！！！！！！！！！！！！")
+            for i in range(3):
+                time.sleep(0.5)
+
+                im_opencv = get_image.get_frame()#[RGB,PIL]
+                detections = {
+                    'center': None,
+                    'next': None,
+                    'npc1': None,
+                    'npc2': None,
+                    'npc3': None,
+                    'npc4': None,
+                    'boss': None,
+                    'daojv': [],
+                    'gw': [],
+                    'zhaozi': None
+
+                }
+                detections = yolo_shibie(im_opencv[1], detections,model)
+                if detections['next'] is not None or len(detections['daojv'])!=0 or len(detections['gw'])!=0 or detections['boss'] is not None:
+                    break
+            else:
+                mouse_gui.send_data_absolute(left + 640, top + 40,may=1)#点击退出
+                panduan = False#退出挑战
+                boss_pd = False
+                time.sleep(2.0)
+                continue
+        if detections['center'] is None and detections['next'] is None and (len(detections['gw'])!=0 or detections["boss"] is not None):#识别不到中心情况 但是有怪物
+            print("未检测到中心点，但是有怪物")
+            if len(detections['gw'])!=0:
+                move_to(rw,detections['gw'][0])
+                time.sleep(0.26)
+            elif detections['boss'] is not None:#跟随boss
+                boss_suiji1=random.randint(-30,30)
+                boss_suiji2 = random.randint(-30, 30)
+                detections['boss']=(detections['boss'][0]+boss_suiji1,detections['boss'][1]+boss_suiji2)
+                move_to(rw,detections['boss'])
+                time.sleep(0.7)
+            continue
+        elif (detections['center'] is not None and detections['next'] is None and len(detections['gw'])!=0) or (boss_pd==True and detections['center'] is not None and detections['next'] is None)    :#识别到中心  但是有怪物
+            if detections['center'][0]>=rw[0] and detections['center'][1]<rw[1]:#3
+                mb=(rw[0]+100,rw[1])
+            elif detections['center'][0]<=rw[0] and detections['center'][1]<rw[1]:#4
+                mb=(rw[0], rw[1]-100)
+            elif detections['center'][0]<=rw[0] and detections['center'][1]>rw[1]:#1
+                mb=(rw[0]-100, rw[1])
+            elif detections['center'][0]>=rw[0] and detections['center'][1]>rw[1]:#2
+                mb=(rw[0], rw[1]+100)
+            move_to(rw, mb)
+            time.sleep(0.25)
+
+
+            continue
+
+        elif boss_pd==True and detections['center'] is None and detections['next'] is None:#boss出现了  但是没有中心
+            k=move_randomly(rw,k)
+            continue
+
+        elif detections['next'] is not None:
+            print('进入下一层啦啦啦啦啦啦')
+            panduan = True
+            move_to(rw,detections['next'])
+            for i in range(2):
+                keyboard.send_data("DD")
+                time.sleep(0.15)
+                keyboard.release()
+            continue
+
+        else:
+            k = move_randomly(rw, k)
+            continue
+    elif shizi.daoying(im_opencv[0]):
+        mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+        time.sleep(1)
+        continue

main_single.py

@@ -0,0 +1,422 @@
+"""
+单配置组运行的主程序
+用于在独立进程中运行单个配置组的自动化
+"""
+import cv2
+from utils.get_image import GetImage
+from utils.mouse import init_mouse_keyboard, Mouse_guiji
+from ultralytics import YOLO
+import time
+import serial
+import ch9329Comm
+import random
+import math
+from utils import shizi
+from config import config_manager
+import sys
+
+def run_automation_for_group(group_index):
+    """为单个配置组运行自动化"""
+    # 重新加载配置
+    config_manager.load_config()
+    group = config_manager.get_group_by_index(group_index)
+    
+    if group is None:
+        print(f"❌ 配置组索引 {group_index} 不存在")
+        return
+    
+    print(f"🚀 启动配置组: {group['name']} (索引: {group_index})")
+    print(f"   串口: {group['serial_port']}")
+    print(f"   采集卡: {group['camera_index']}")
+    
+    # 加载YOLO模型
+    model = YOLO(r"best.pt").to('cuda')
+    model0 = YOLO(r"best0.pt").to('cuda')
+    
+    # 初始化串口和鼠标
+    try:
+        init_mouse_keyboard(group)
+    except Exception as e:
+        print(f"❌ 初始化串口失败: {e}")
+        return
+    
+    # 初始化键盘和鼠标
+    keyboard = ch9329Comm.keyboard.DataComm()
+    from utils.mouse import mouse
+    mouse_gui = Mouse_guiji()
+    
+    # 初始化采集卡
+    get_image = GetImage(
+        cam_index=group['camera_index'],
+        width=group['camera_width'],
+        height=group['camera_height']
+    )
+    
+    # 检查采集卡是否初始化成功
+    if get_image.cap is None:
+        print(f"❌ 采集卡 {group['camera_index']} 初始化失败")
+        return
+    
+    print(f"✅ 配置组 {group['name']} 初始化完成")
+    
+    # 全局变量
+    left = 0
+    top = 30
+    k = 0  # 控制转圈的方向
+    panduan = False  # 是否在图内
+    boss_pd = False  # 是否到boss关卡
+    rw = (632, 378)
+    v = group['move_velocity']  # 从配置读取移动速度
+    
+    def yolo_shibie(im_PIL, detections, model):
+        try:
+            results = model(im_PIL)
+            for result in results:
+                if result.boxes is None or len(result.boxes.xyxy) == 0:
+                    continue
+                for i in range(len(result.boxes.xyxy)):
+                    try:
+                        left = float(result.boxes.xyxy[i][0])
+                        top = float(result.boxes.xyxy[i][1])
+                        right = float(result.boxes.xyxy[i][2])
+                        bottom = float(result.boxes.xyxy[i][3])
+                        cls_id = int(result.boxes.cls[i])
+                        label = result.names[cls_id]
+                        
+                        if label == 'center' or label == 'next' or label == 'boss' or label == 'zhaozi':
+                            player_x = int(left + (right - left) / 2) + 3
+                            player_y = int(top + (bottom - top) / 2) + 40
+                            RW = [player_x, player_y]
+                            detections[label] = RW
+                        elif label == 'daojv' or label == 'gw':
+                            player_x = int(left + (right - left) / 2) + 3
+                            player_y = int(top + (bottom - top) / 2) + 40
+                            RW = [player_x, player_y]
+                            if label not in detections:
+                                detections[label] = []
+                            detections[label].append(RW)
+                        elif label == 'npc1' or label == 'npc2' or label == 'npc3' or label == 'npc4':
+                            player_x = int(left + (right - left) / 2)
+                            player_y = int(bottom) + 30
+                            RW = [player_x, player_y]
+                            detections[label] = RW
+                    except Exception as e:
+                        print(f"⚠️ 处理检测框时出错: {e}")
+                        continue
+        except Exception as e:
+            print(f"⚠️ YOLO检测出错: {e}")
+        return detections
+    
+    def sq(p1, p2):
+        """计算两点之间的欧式距离"""
+        return math.sqrt((p1[0] - p2[0]) ** 2 + (p1[1] - p2[1]) ** 2)
+    
+    def process_points(points):
+        if not points:
+            return None
+        
+        n = len(points)
+        if n == 1:
+            return points[0]
+        elif n == 2:
+            x = (points[0][0] + points[1][0]) / 2
+            y = (points[0][1] + points[1][1]) / 2
+            return [x, y]
+        else:
+            sample_points = random.sample(points, 3)
+            min_sum = float('inf')
+            best_point = None
+            for p in points:
+                dist_sum = sum(sq(p, sp) for sp in sample_points)
+                if dist_sum < min_sum:
+                    min_sum = dist_sum
+                    best_point = p
+            return best_point
+    
+    def safe_keyboard_send(data, max_retries=3):
+        """安全的键盘发送函数，带重试机制"""
+        nonlocal keyboard
+        for attempt in range(max_retries):
+            try:
+                keyboard.send_data(data)
+                return True
+            except (serial.serialutil.SerialException, PermissionError, OSError) as e:
+                if attempt < max_retries - 1:
+                    print(f"⚠️ 键盘发送失败 (尝试 {attempt + 1}/{max_retries}): {e}")
+                    time.sleep(0.1 * (attempt + 1))  # 递增延迟
+                    # 尝试重新初始化串口
+                    try:
+                        if serial.ser and serial.ser.is_open:
+                            serial.ser.close()
+                        time.sleep(0.2)
+                        init_mouse_keyboard(group)
+                        # 重新创建keyboard对象
+                        keyboard = ch9329Comm.keyboard.DataComm()
+                    except Exception as init_e:
+                        print(f"⚠️ 重新初始化串口失败: {init_e}")
+                else:
+                    print(f"❌ 键盘发送失败，已重试 {max_retries} 次: {e}")
+                    raise
+        return False
+    
+    def safe_keyboard_release(max_retries=3):
+        """安全的键盘释放函数，带重试机制"""
+        nonlocal keyboard
+        for attempt in range(max_retries):
+            try:
+                keyboard.release()
+                return True
+            except (serial.serialutil.SerialException, PermissionError, OSError) as e:
+                if attempt < max_retries - 1:
+                    time.sleep(0.1 * (attempt + 1))
+                    # 尝试重新初始化串口
+                    try:
+                        if serial.ser and serial.ser.is_open:
+                            serial.ser.close()
+                        time.sleep(0.2)
+                        init_mouse_keyboard(group)
+                        keyboard = ch9329Comm.keyboard.DataComm()
+                    except Exception as init_e:
+                        print(f"⚠️ 重新初始化串口失败: {init_e}")
+                else:
+                    print(f"⚠️ 键盘释放失败: {e}")
+                    # 释放失败不算致命错误，继续执行
+                    return False
+        return False
+    
+    def move_randomly(rw, k):
+        k = k % 4
+        suiji_t = float(random.randint(10, 13) / 10)
+        if k == 0:
+            if safe_keyboard_send("66"):
+                time.sleep(suiji_t)
+                safe_keyboard_release()
+        elif k == 1:
+            if safe_keyboard_send("88"):
+                time.sleep(suiji_t)
+                safe_keyboard_release()
+        elif k == 2:
+            if safe_keyboard_send("44"):
+                time.sleep(suiji_t)
+                safe_keyboard_release()
+        elif k == 3:
+            if safe_keyboard_send("22"):
+                time.sleep(suiji_t)
+                safe_keyboard_release()
+        return k + 1
+    
+    def move_to(rw, mb):
+        """使用配置的v值"""
+        nonlocal v
+        v = group['move_velocity']
+        if rw[0] >= mb[0]:
+            if safe_keyboard_send("44"):
+                time.sleep(float(abs(rw[0] - mb[0]) / v))
+                safe_keyboard_release()
+        else:
+            if safe_keyboard_send("66"):
+                time.sleep(float(abs(rw[0] - mb[0]) / v))
+                safe_keyboard_release()
+        
+        if rw[1] >= mb[1]:
+            if safe_keyboard_send("88"):
+                time.sleep(float(abs(rw[1] - mb[1]) / v))
+                safe_keyboard_release()
+        else:
+            if safe_keyboard_send("22"):
+                time.sleep(float(abs(rw[1] - mb[1]) / v))
+                safe_keyboard_release()
+    
+    # 主循环
+    print(f"🔄 配置组 {group['name']} 开始自动化循环...")
+    while True:
+        try:
+            detections = {
+                'center': None,
+                'next': None,
+                'npc1': None,
+                'npc2': None,
+                'npc3': None,
+                'npc4': None,
+                'boss': None,
+                'daojv': [],
+                'gw': [],
+                'zhaozi': None
+            }
+            im_opencv = get_image.get_frame()
+            if im_opencv is None:
+                time.sleep(0.1)
+                continue
+            
+            detections = yolo_shibie(im_opencv[1], detections, model)
+            
+            if shizi.tuwai(im_opencv[0]):
+                im_opencv = get_image.get_frame()
+                if im_opencv is None:
+                    time.sleep(0.1)
+                    continue
+                detections = yolo_shibie(im_opencv[1], detections, model0)
+                print(f'[{group["name"]}] 当前在城镇中')
+                if detections['npc1'] is not None and sq(rw, detections['npc1']) > 80:
+                    print(f"[{group['name']}] 向npc1靠近")
+                    move_to(rw, detections['npc1'])
+                    continue
+                elif detections['npc1'] is not None and sq(rw, detections['npc1']) <= 80:
+                    print(f"[{group['name']}] 在npc旁边，向上走")
+                    mb = (detections['npc1'][0], detections['npc1'][1] - 1010)
+                    move_to(rw, mb)
+                    continue
+                elif detections['npc3'] is not None and detections['npc4'] is None:
+                    print(f"[{group['name']}] 在npc3旁边，向右走")
+                    mb = (rw[0], detections['npc3'][1] - 50)
+                    move_to(rw, mb)
+                    mb = (rw[0] + 700, rw[1])
+                    move_to(rw, mb)
+                    continue
+                elif detections['npc4'] is not None:
+                    if sq(detections['npc4'], rw) < 50:
+                        print(f"[{group['name']}] 离npc4很近 直接进入")
+                        if safe_keyboard_send("DD"):
+                            time.sleep(0.15)
+                            safe_keyboard_release()
+                        time.sleep(1)
+                        im_opencv = get_image.get_frame()
+                        if im_opencv and shizi.daoying(im_opencv[0]):
+                            mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                            time.sleep(1)
+                        continue
+                    else:
+                        print(f"[{group['name']}] 离npc4有点远 点击进入")
+                        move_to(rw, detections['npc4'])
+                        time.sleep(1)
+                        im_opencv = get_image.get_frame()
+                        if im_opencv and shizi.daoying(im_opencv[0]):
+                            mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                            time.sleep(1)
+                        continue
+            elif shizi.tiaozhan(im_opencv[0]):
+                print(f'[{group["name"]}] 进入塔4')
+                mouse_gui.send_data_absolute(left + 1100, top + 600, may=1)
+                time.sleep(0.3)
+                mouse_gui.send_data_absolute(left + 433, top + 455, may=1)
+                panduan = True
+                continue
+            elif shizi.jieshu(im_opencv[0]):
+                print(f'[{group["name"]}] 结束挑战')
+                mouse_gui.send_data_absolute(left + 542, top + 644, may=1)
+                time.sleep(0.8)
+                mouse_gui.send_data_absolute(left + 706, top + 454, may=1)
+                continue
+            elif panduan:
+                if shizi.shuzi(im_opencv[0]):
+                    boss_pd = True
+                    print(f"[{group['name']}] 进入到boss！！！！！！！！！！！！！！！！！！")
+                if shizi.fuhuo(im_opencv[0]):
+                    print(f'[{group["name"]}] 点击复活')
+                    mouse_gui.send_data_absolute(left + 536, top + 627, may=1)
+                    mouse_gui.send_data_absolute(rw[0], rw[1], may=0)
+                    continue
+                if detections['zhaozi'] is not None:
+                    move_to(rw, detections['zhaozi'])
+                    continue
+                if len(detections['daojv']) != 0:
+                    move_to(rw, process_points(detections['daojv']))
+                    for i in range(3 + len(detections['daojv'])):
+                        if safe_keyboard_send("AA"):
+                            time.sleep(0.15)
+                            safe_keyboard_release()
+                    continue
+                if shizi.tuichu(im_opencv[0]) and detections['next'] is None and len(detections['daojv']) == 0 and len(detections['gw']) == 0 and boss_pd:
+                    print(f"[{group['name']}] 识别到可以退出挑战！！！！！！！！！！！！！！！！！！")
+                    for i in range(3):
+                        time.sleep(0.5)
+                        im_opencv = get_image.get_frame()
+                        if im_opencv is None:
+                            continue
+                        detections = {
+                            'center': None, 'next': None,
+                            'npc1': None, 'npc2': None, 'npc3': None, 'npc4': None,
+                            'boss': None, 'zhaozi': None,
+                            'daojv': [], 'gw': []
+                        }
+                        detections = yolo_shibie(im_opencv[1], detections, model)
+                        if detections['next'] is not None or len(detections['daojv']) != 0 or len(detections['gw']) != 0 or detections['boss'] is not None:
+                            break
+                    else:
+                        mouse_gui.send_data_absolute(left + 640, top + 40, may=1)
+                        panduan = False
+                        boss_pd = False
+                        time.sleep(2.0)
+                        continue
+                if detections['center'] is None and detections['next'] is None and (len(detections['gw']) != 0 or detections["boss"] is not None):
+                    if len(detections['gw']) != 0:
+                        move_to(rw, detections['gw'][0])
+                        time.sleep(0.26)
+                    elif detections['boss'] is not None:
+                        boss_suiji1 = random.randint(-30, 30)
+                        boss_suiji2 = random.randint(-30, 30)
+                        detections['boss'] = (detections['boss'][0] + boss_suiji1, detections['boss'][1] + boss_suiji2)
+                        move_to(rw, detections['boss'])
+                        time.sleep(0.7)
+                    continue
+                elif (detections['center'] is not None and detections['next'] is None and len(detections['gw']) != 0) or (boss_pd == True and detections['center'] is not None and detections['next'] is None):
+                    if detections['center'][0] >= rw[0] and detections['center'][1] < rw[1]:
+                        mb = (rw[0] + 100, rw[1])
+                    elif detections['center'][0] <= rw[0] and detections['center'][1] < rw[1]:
+                        mb = (rw[0], rw[1] - 100)
+                    elif detections['center'][0] <= rw[0] and detections['center'][1] > rw[1]:
+                        mb = (rw[0] - 100, rw[1])
+                    elif detections['center'][0] >= rw[0] and detections['center'][1] > rw[1]:
+                        mb = (rw[0], rw[1] + 100)
+                    move_to(rw, mb)
+                    time.sleep(0.25)
+                    continue
+                elif boss_pd == True and detections['center'] is None and detections['next'] is None:
+                    k = move_randomly(rw, k)
+                    continue
+                elif detections['next'] is not None:
+                    print(f'[{group["name"]}] 进入下一层啦啦啦啦啦啦')
+                    panduan = True
+                    move_to(rw, detections['next'])
+                    for i in range(2):
+                        if safe_keyboard_send("DD"):
+                            time.sleep(0.15)
+                            safe_keyboard_release()
+                    continue
+                else:
+                    k = move_randomly(rw, k)
+                    continue
+            elif shizi.daoying(im_opencv[0]):
+                mouse_gui.send_data_absolute(rw[0], rw[1] - 110, may=1)
+                time.sleep(1)
+                continue
+        except KeyboardInterrupt:
+            print(f"🛑 配置组 {group['name']} 停止运行")
+            break
+        except Exception as e:
+            print(f"❌ 配置组 {group['name']} 运行错误: {e}")
+            import traceback
+            traceback.print_exc()
+            time.sleep(1)
+    
+    # 清理资源
+    try:
+        if serial.ser and serial.ser.is_open:
+            serial.ser.close()
+            print(f"✅ 配置组 {group['name']} 串口已关闭")
+    except Exception as e:
+        print(f"⚠️ 关闭串口时出错: {e}")
+
+if __name__ == "__main__":
+    if len(sys.argv) > 1:
+        group_index = int(sys.argv[1])
+        run_automation_for_group(group_index)
+    else:
+        # 默认运行活动配置组
+        active_group = config_manager.get_active_group()
+        if active_group is None:
+            print("❌ 没有活动的配置组")
+            sys.exit(1)
+        group_index = config_manager.config['groups'].index(active_group)
+        run_automation_for_group(group_index)
+

preview.py

@@ -0,0 +1,586 @@
+import cv2
+import tkinter as tk
+from tkinter import Canvas
+from PIL import Image, ImageTk
+import threading
+import numpy as np
+import warnings
+import os
+from config import config_manager
+
+# 抑制OpenCV的警告信息（兼容不同版本）
+import sys
+import io
+import time as _time
+
+os.environ['OPENCV_LOG_LEVEL'] = 'SILENT'
+os.environ['OPENCV_IO_ENABLE_OPENEXR'] = '0'
+
+try:
+    if hasattr(cv2, 'setLogLevel'):
+        if hasattr(cv2, 'LOG_LEVEL_SILENT'):
+            cv2.setLogLevel(cv2.LOG_LEVEL_SILENT)
+        elif hasattr(cv2, 'LOG_LEVEL_ERROR'):
+            cv2.setLogLevel(cv2.LOG_LEVEL_ERROR)
+        elif hasattr(cv2, 'utils'):
+            cv2.utils.setLogLevel(0)
+except Exception:
+    pass
+
+# 简单日志限流
+_last_log_times = {}
+def log_throttle(key: str, interval_sec: float, text: str):
+    now = _time.time()
+    last = _last_log_times.get(key, 0)
+    if now - last >= interval_sec:
+        _last_log_times[key] = now
+        try:
+            print(text)
+        except Exception:
+            pass
+
+class PreviewWindow:
+    """采集卡预览窗口"""
+    def __init__(self):
+        self.config = config_manager.config
+        self.caps = {}
+        self.frames = {}
+        self.large_windows = {}  # idx -> Toplevel
+        self.running = True
+        
+    def init_cameras(self):
+        """初始化所有相机"""
+        print("🔧 开始初始化采集卡...")
+        loaded_count = 0
+        
+        # 根据配置选择加载哪些组
+        display = self.config.get('display', {})
+        use_all = bool(display.get('preview_use_all_groups', True))
+        if use_all:
+            target_groups = self.config['groups']
+            print(f"📷 预览全部配置组，共 {len(target_groups)} 组")
+        else:
+            target_groups = [g for g in self.config['groups'] if g.get('active', False)]
+            print(f"📷 仅预览活动配置组，共 {len(target_groups)} 组")
+            if not target_groups:
+                print("⚠️ 没有活动的配置组，将不显示任何采集卡。可在配置中启用‘预览全部配置组’或设置活动组")
+        
+        # 重定向stderr来抑制OpenCV的错误输出
+        old_stderr = sys.stderr
+        suppressed_output = io.StringIO()
+        
+        try:
+            sys.stderr = suppressed_output
+            
+            for i, group in enumerate(target_groups):
+                try:
+                    cam_idx = group['camera_index']
+                    print(f"  尝试打开采集卡 {cam_idx} ({group['name']})...")
+                    
+                    cap = None
+                    # 尝试多种后端打开
+                    backends_to_try = [
+                        (int(cam_idx), cv2.CAP_DSHOW),
+                        (int(cam_idx), cv2.CAP_ANY),
+                        (int(cam_idx), None),
+                    ]
+                    
+                    for idx, backend in backends_to_try:
+                        try:
+                            with warnings.catch_warnings():
+                                warnings.filterwarnings('ignore')
+                                if backend is not None:
+                                    cap = cv2.VideoCapture(idx, backend)
+                                else:
+                                    cap = cv2.VideoCapture(idx)
+                            
+                            if cap.isOpened():
+                                # 测试读取一帧
+                                ret, test_frame = cap.read()
+                                if ret and test_frame is not None:
+                                    break
+                                else:
+                                    cap.release()
+                                    cap = None
+                        except Exception:
+                            if cap:
+                                try:
+                                    cap.release()
+                                except:
+                                    pass
+                                cap = None
+                            continue
+                    
+                    if cap and cap.isOpened():
+                        try:
+                            cap.set(cv2.CAP_PROP_FRAME_WIDTH, group['camera_width'])
+                            cap.set(cv2.CAP_PROP_FRAME_HEIGHT, group['camera_height'])
+                        except Exception as e:
+                            print(f"    ⚠️ 设置分辨率失败: {e}")
+                        
+                        self.caps[i] = {
+                            'cap': cap,
+                            'group': group,
+                            'name': group['name']
+                        }
+                        loaded_count += 1
+                        print(f"  ✅ 采集卡 {cam_idx} 初始化成功")
+                    else:
+                        print(f"  ❌ 采集卡 {cam_idx} 无法打开")
+                except Exception as e:
+                    print(f"  ❌ 采集卡 {group.get('camera_index', '?')} 初始化失败: {e}")
+                    import traceback
+                    traceback.print_exc()
+        finally:
+            # 恢复stderr
+            sys.stderr = old_stderr
+        
+        if loaded_count == 0:
+            print("⚠️ 警告：没有成功加载任何采集卡！")
+            print("请检查：")
+            print("1. 采集卡是否正确连接")
+            print("2. 采集卡索引是否正确")
+            print("3. 是否有活动的配置组")
+        else:
+            print(f"✅ 成功加载 {loaded_count} 个采集卡")
+    
+    def grab_once(self):
+        """抓取每個采集卡當前單幀（裁剪後，存入 self.frames），抓取结果增强debug输出"""
+        for idx, data in self.caps.items():
+            try:
+                cap = data['cap']
+                frm_name = data.get('name', str(idx))
+                if cap is None:
+                    print(f"[抓取] 采集卡{idx}({frm_name}) cap为None")
+                    self.frames[idx] = None
+                    continue
+                ret, frame = cap.read()
+                if not ret or frame is None:
+                    print(f"[抓取] 采集卡{idx}({frm_name}) 没抓到帧 ret={ret} frame=None")
+                    self.frames[idx] = None
+                    continue
+                height, width = frame.shape[:2]
+                crop_top = 30
+                crop_bottom = min(crop_top + 720, height)
+                crop_width = min(1280, width)
+                if crop_bottom <= crop_top or crop_width <= 0:
+                    print(f"[抓取] 采集卡{idx}({frm_name}) 尺寸异常: H={height} W={width} 裁剪失败")
+                    self.frames[idx] = None
+                    continue
+                frame = frame[crop_top:crop_bottom, 0:crop_width]
+                # 额外检查：像素全黑？
+                if np.mean(frame) < 3:
+                    print(f"[抓取] 采集卡{idx}({frm_name}) 帧抓到但内容接近全黑 shape={frame.shape}")
+                else:
+                    print(f"[抓取] 采集卡{idx}({frm_name}) 成功 shape={frame.shape} 类型={frame.dtype}")
+                self.frames[idx] = frame
+            except Exception as e:
+                print(f"[抓取] 采集卡{idx}({data.get('name', str(idx))}) 抓取异常: {e}")
+                self.frames[idx] = None
+    
+    def capture_frames(self):
+        """捕获帧"""
+        while self.running:
+            for idx, data in self.caps.items():
+                try:
+                    ret, frame = data['cap'].read()
+                    if ret and frame is not None:
+                        # 裁剪到配置的区域
+                        height, width = frame.shape[:2]
+                        crop_top = 30
+                        crop_bottom = min(crop_top + 720, height)
+                        crop_width = min(1280, width)
+                        
+                        # 确保裁剪范围有效
+                        if crop_bottom > crop_top and crop_width > 0:
+                            frame = frame[crop_top:crop_bottom, 0:crop_width]
+                            
+                            # 转换颜色空间
+                            frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+                            self.frames[idx] = frame_rgb
+                        else:
+                            print(f"⚠️ 采集卡 {idx} 裁剪参数无效")
+                    else:
+                        # 读取失败，清除旧帧
+                        if idx in self.frames:
+                            self.frames[idx] = None
+                except Exception as e:
+                    print(f"捕获帧 {idx} 错误: {e}")
+                    import traceback
+                    traceback.print_exc()
+            
+            import time
+            time.sleep(0.01)  # 避免CPU占用过高
+    
+    def create_grid_window(self, master=None):
+        """创建网格窗口，可挂载在主Tk下作为Toplevel弹窗"""
+        display = self.config.get('display', {})
+        preview_width = display.get('preview_width', 1000)
+        preview_height = display.get('preview_height', 700)
+        columns = display.get('preview_columns', 2)
+        rows = display.get('preview_rows', 2)
+
+        if master is not None:
+            root = tk.Toplevel(master)
+        else:
+            root = tk.Tk()
+        root.title("采集卡预览 - 点击放大")
+        root.geometry(f"{preview_width}x{preview_height}")
+        root.update_idletasks()
+        
+        canvas = Canvas(root, bg='black', width=preview_width, height=preview_height)
+        canvas.pack(fill=tk.BOTH, expand=True)
+        
+        # 存储图像对象
+        self.photo_objects = {}
+        
+        # 用于控制调试输出（只打印前几次）
+        self.debug_count = 0
+        
+        def update_frames_once():
+            """在主线程中每5秒刷新画面（直接显示高频采集线程生成的帧，去除grab_once）"""
+            if not self.running:
+                return
+            try:
+                if not self.caps:
+                    canvas.delete("all")
+                    canvas.create_text(
+                        root.winfo_width() // 2, 
+                        root.winfo_height() // 2, 
+                        text="未找到可用的采集卡\n\n请检查：\n1. 采集卡是否正确连接\n2. 配置组是否有活动的采集卡\n3. 采集卡索引是否正确", 
+                        fill='yellow', 
+                        font=('Arial', 14),
+                        justify=tk.CENTER
+                    )
+                    root.after(5000, update_frames_once)
+                    return
+
+                # 计算每个预览窗口的位置和大小
+                # 直接使用配置值作为画布尺寸（macOS上窗口尺寸可能在显示前返回默认值）
+                canvas_width = preview_width
+                canvas_height = preview_height
+                
+                # 尝试获取实际的窗口尺寸，如果有效则使用（大于配置值说明可能被手动调整了）
+                try:
+                    root.update_idletasks()
+                    actual_width = root.winfo_width()
+                    actual_height = root.winfo_height()
+                    
+                    # 只有在获取到合理的尺寸时才使用（大于100像素）
+                    if actual_width > 100 and actual_height > 100:
+                        canvas_width = actual_width
+                        canvas_height = actual_height
+                except:
+                    pass  # 如果获取失败，使用配置值
+                
+                cell_width = max(10, canvas_width // columns)
+                cell_height = max(10, canvas_height // rows)
+
+                # 先收集所有需要显示的图像
+                images_to_draw = []
+                texts_to_draw = []
+                frame_idx = 0
+                
+                for idx in self.caps.keys():
+                    frame = self.frames.get(idx)
+                    has_frame = frame is not None
+                    print(f'[DEBUG] update UI idx={idx}, has_frame={has_frame}, frame_info={getattr(frame, "shape", None)}')
+                    name = self.caps[idx]['name']
+                    row = frame_idx // columns
+                    col = frame_idx % columns
+                    x = col * cell_width
+                    y = row * cell_height
+                    center_x = x + cell_width // 2
+                    center_y = y + cell_height // 2
+                    # 如果没拿到frame，临时主动再抓一次，并DEBUG
+                    if not has_frame:
+                        print(f'[DEBUG] UI未获得frame，尝试临时抓取采集卡{idx}...')
+                        try:
+                            cap_test = self.caps[idx]['cap']
+                            ret, fr2 = cap_test.read()
+                            print(f'[DEBUG] 临时抓取ret={ret}, shape={getattr(fr2, "shape", None)}')
+                            if ret and fr2 is not None:
+                                frame = fr2
+                                has_frame = True
+                            else:
+                                frame = None
+                                has_frame = False
+                        except Exception as e:
+                            print(f'[DEBUG] 临时抓取异常：{e}')
+                            frame = None
+                            has_frame = False
+                    if has_frame:
+                        try:
+                            h, w = frame.shape[:2]
+                            if w <= 0 or h <= 0:
+                                detail = f"采集卡{idx}:{name}\n抓到帧但图像尺寸异常 ({w}x{h})"
+                                texts_to_draw.append((center_x, center_y, detail, 'red'))
+                                frame_idx += 1
+                                continue
+                            scale = min(cell_width / w, cell_height / h, 1.0) * 0.85
+                            new_w = max(1, int(w * scale))
+                            new_h = max(1, int(h * scale))
+                            if new_w <= 0 or new_h <= 0:
+                                detail = f"采集卡{idx}:{name}\n缩放计算异常 ({w}x{h})"
+                                texts_to_draw.append((center_x, center_y, detail, 'red'))
+                                frame_idx += 1
+                                continue
+                            display_frame = frame
+                            if len(display_frame.shape) == 3 and display_frame.shape[2] == 3:
+                                display_frame = cv2.cvtColor(display_frame, cv2.COLOR_BGR2RGB)
+                            resized_frame = cv2.resize(display_frame, (new_w, new_h))
+                            pil_image = Image.fromarray(resized_frame)
+                            photo = ImageTk.PhotoImage(image=pil_image)
+                            if not hasattr(self, 'photo_refs'):
+                                self.photo_refs = {}
+                            self.photo_refs[idx] = photo
+                            canvas.create_image(center_x, center_y, image=photo, anchor='center')
+                            canvas.update()  # 保证实时刷新
+                            print(f'[DEBUG] UI idx={idx}, 绘制photo id={id(photo)}, photo_refs={len(self.photo_refs)}')
+                            images_to_draw.append((photo, center_x, center_y))
+                            texts_to_draw.append((center_x, y + 15, name, 'white'))
+                            frame_idx += 1
+                        except Exception as e:
+                            errstr = str(e)
+                            detail = f"采集卡{idx}:{name}\n帧处理异常：{errstr}"
+                            print(f'[DEBUG] idx={idx}, 帧处理异常：{errstr}')
+                            texts_to_draw.append((center_x, center_y, detail, 'red'))
+                            frame_idx += 1
+                    else:
+                        detail = f"采集卡{idx}:{name}\n抓取失败/暂无画面"
+                        print(f'[DEBUG] idx={idx} 依然无可用帧，最终显示错误提示')
+                        texts_to_draw.append((center_x, center_y, detail, 'red'))
+                        frame_idx += 1
+                
+                # 清空画布
+                canvas.delete("all")
+                
+                # 先绘制所有图像（底层）
+                if images_to_draw and self.debug_count <= 3:
+                    log_throttle("draw_prepare", 2.0, f"✅ 准备绘制 {len(images_to_draw)} 个图像到画布 ({canvas_width}x{canvas_height})")
+                    
+                for photo, x, y in images_to_draw:
+                    try:
+                        # PhotoImage强引用，防止被GC
+                        if not hasattr(self, 'photo_refs'):
+                            self.photo_refs = {}
+                        self.photo_refs[idx] = photo
+                        canvas.create_image(x, y, image=photo, anchor='center')
+                        canvas.update()  # 保证实时刷新
+                        print(f'[DEBUG] idx={idx}, photo id={id(photo)}, photo refs={len(self.photo_refs)}')
+                    except Exception as e:
+                        if "pyimage" not in str(e).lower():
+                            log_throttle("draw_image_error", 2.0, f"绘制图像错误: {e}")
+                
+                # 再绘制所有文本（上层）
+                for x, y, text, color in texts_to_draw:
+                    try:
+                        if color == 'white':
+                            canvas.create_text(x, y, text=text, fill=color, font=('Arial', 10, 'bold'))
+                        else:
+                            canvas.create_text(x, y, text=text, fill=color, font=('Arial', 10))
+                    except Exception as e:
+                        log_throttle("draw_text_error", 2.0, f"绘制文本错误: {e}")
+                
+                # 绘制分割线，区分不同的采集卡窗口
+                try:
+                    # 绘制垂直分割线（列之间的分割线）
+                    for col in range(1, columns):
+                        x = col * cell_width
+                        canvas.create_line(
+                            x, 0, 
+                            x, canvas_height,
+                            fill='white', 
+                            width=2,
+                            dash=(5, 5)  # 虚线效果，让分割线更明显
+                        )
+                    
+                    # 绘制水平分割线（行之间的分割线）
+                    for row in range(1, rows):
+                        y = row * cell_height
+                        canvas.create_line(
+                            0, y,
+                            canvas_width, y,
+                            fill='white',
+                            width=2,
+                            dash=(5, 5)  # 虚线效果
+                        )
+                except Exception as e:
+                    log_throttle("draw_grid_error", 2.0, f"绘制分割线错误: {e}")
+                        
+            except Exception as e:
+                log_throttle("update_frame_error", 1.0, f"更新帧错误: {e}")
+                import traceback
+                traceback.print_exc()
+            # 每5秒更新一次
+            root.after(5000, update_frames_once)
+        
+        def on_canvas_click(event):
+            """点击画布事件"""
+            window_width = root.winfo_width()
+            window_height = root.winfo_height()
+            cell_width = window_width // columns
+            cell_height = window_height // rows
+            
+            col = int(event.x // cell_width)
+            row = int(event.y // cell_height)
+            index = row * columns + col
+            
+            # 找到对应的配置
+            if index < len(self.caps):
+                idx = list(self.caps.keys())[index]
+                self.show_large_window(idx)
+        
+        canvas.bind('<Button-1>', on_canvas_click)
+        
+        # 等待窗口完全初始化后再开始更新
+        def start_updates():
+            """延迟启动更新，确保窗口已完全显示"""
+            # 强制更新窗口尺寸
+            root.update_idletasks()
+            root.update()
+            # 等待窗口完全绘制
+            import time
+            time.sleep(0.2)  # 给窗口更多时间初始化
+            root.update_idletasks()
+            root.update()
+            update_frames_once()
+        
+        # 使用after在主线程中循环刷新（延迟启动，给足够时间让窗口初始化）
+        root.after(200, start_updates)
+        
+        def on_closing():
+            """关闭窗口"""
+            self.running = False
+            for data in self.caps.values():
+                data['cap'].release()
+            root.destroy()
+        
+        root.protocol("WM_DELETE_WINDOW", on_closing)
+        # 仅在独立窗口（无master）时进入事件循环
+        if master is None:
+            root.mainloop()
+    
+    def show_large_window(self, idx):
+        """显示指定采集卡的大窗口（可多实例）"""
+        # 如果已存在该索引窗口，先销毁再创建，确保刷新
+        if idx in self.large_windows and self.large_windows[idx].winfo_exists():
+            try:
+                self.large_windows[idx].destroy()
+            except Exception:
+                pass
+        
+        win = tk.Toplevel()
+        win.title(f"放大视图 - {self.caps[idx]['name']}")
+        win.geometry("1280x720")
+        self.large_windows[idx] = win
+        
+        canvas = Canvas(win, bg='black')
+        canvas.pack(fill=tk.BOTH, expand=True)
+        
+        photo_obj = {}
+        
+        def update_large_once():
+            if not self.running or not win.winfo_exists():
+                return
+            try:
+                window_width = win.winfo_width() if win.winfo_width() > 1 else 1280
+                window_height = win.winfo_height() if win.winfo_height() > 1 else 720
+                
+                if idx in self.frames and self.frames[idx] is not None:
+                    frame = self.frames[idx]
+                    h, w = frame.shape[:2]
+                    scale = min(window_width / w, window_height / h)
+                    new_w = int(w * scale)
+                    new_h = int(h * scale)
+                    resized_frame = cv2.resize(frame, (new_w, new_h))
+                    if len(resized_frame.shape) == 3 and resized_frame.shape[2] == 3:
+                        resized_frame = cv2.cvtColor(resized_frame, cv2.COLOR_BGR2RGB)
+                    pil_image = Image.fromarray(resized_frame)
+                    photo = ImageTk.PhotoImage(image=pil_image)
+                    photo_obj['img'] = photo
+                    canvas.delete("all")
+                    canvas.create_image(window_width // 2, window_height // 2, image=photo, anchor='center')
+                else:
+                    canvas.delete("all")
+                    # 显示更详细的调试信息
+                    debug_text = f"等待画面...\n采集卡: {idx}\n帧状态: {idx in self.frames}\n"
+                    if idx in self.frames:
+                        debug_text += f"帧数据: {self.frames[idx] is not None}"
+                    else:
+                        debug_text += "帧数据: 未初始化"
+                    canvas.create_text(
+                        window_width // 2,
+                        window_height // 2,
+                        text=debug_text,
+                        fill='gray',
+                        font=('Arial', 12),
+                        justify='center'
+                    )
+            except Exception as e:
+                if "pyimage" not in str(e).lower():
+                    log_throttle("large_update_error", 1.0, f"更新大窗口错误: {e}")
+            # 放大視窗維持原本高頻刷新
+            win.after(33, update_large_once)
+
+        win.after(33, update_large_once)
+
+        def on_close():
+            try:
+                if idx in self.large_windows:
+                    del self.large_windows[idx]
+            finally:
+                win.destroy()
+        win.protocol("WM_DELETE_WINDOW", on_close)
+    
+    def run(self, master=None):
+        """运行预览，支持主Tk弹窗模式"""
+        self.init_cameras()
+
+        if not self.caps:
+            print("⚠️ 没有可用的采集卡，将显示错误提示窗口")
+
+        display = self.config.get('display', {})
+        multi_window = bool(display.get('preview_multi_window', False))
+
+        if self.caps:
+            capture_thread = threading.Thread(target=self.capture_frames, daemon=True)
+            capture_thread.start()
+            import time
+            time.sleep(0.5)
+
+        if multi_window:
+            root = tk.Toplevel(master) if master else tk.Tk()
+            root.title("采集卡预览 - 多窗口模式")
+            root.geometry("300x80")
+            tk.Label(root, text="已打开多窗口预览，可单独拖动/调整大小").pack(pady=20)
+            print("等待采集卡初始化...")
+            max_wait = 50
+            wait_count = 0
+            while wait_count < max_wait and not any(frame is not None for frame in self.frames.values()):
+                time.sleep(0.1)
+                wait_count += 1
+            if wait_count >= max_wait:
+                print("⚠️ 采集卡初始化超时，可能无法显示画面")
+            for idx in list(self.caps.keys()):
+                print(f"打开采集卡 {idx} 的预览窗口...")
+                self.show_large_window(idx)
+            def on_root_close():
+                self.running = False
+                for data in self.caps.values():
+                    data['cap'].release()
+                try:
+                    root.destroy()
+                except Exception:
+                    pass
+            root.protocol("WM_DELETE_WINDOW", on_root_close)
+            # 仅在独立窗口（无master）时进入事件循环
+            if master is None:
+                root.mainloop()
+        else:
+            self.create_grid_window(master)
+
+if __name__ == "__main__":
+    preview = PreviewWindow()
+    preview.run()
+

test_camera.py

@@ -0,0 +1,67 @@
+"""
+采集卡测试脚本
+用于测试采集卡是否能正常工作
+"""
+import cv2
+from utils.get_image import GetImage
+from config import config_manager
+
+def test_camera():
+    """测试采集卡"""
+    print("🔧 采集卡测试工具")
+    print("=" * 50)
+    
+    # 获取活动配置
+    active_group = config_manager.get_active_group()
+    if active_group is None:
+        print("❌ 没有活动的配置组")
+        print("请先运行 python gui_config.py 设置配置")
+        return
+    
+    print(f"📋 使用配置: {active_group['name']}")
+    print(f"   采集卡索引: {active_group['camera_index']}")
+    print(f"   分辨率: {active_group['camera_width']}x{active_group['camera_height']}")
+    print()
+    
+    # 初始化采集卡
+    get_image = GetImage(
+        cam_index=active_group['camera_index'],
+        width=active_group['camera_width'],
+        height=active_group['camera_height']
+    )
+    
+    if get_image.cap is None:
+        print("❌ 采集卡初始化失败")
+        return
+    
+    print("✅ 采集卡初始化成功")
+    print("按 'q' 退出测试")
+    print()
+    
+    # 测试循环
+    frame_count = 0
+    while True:
+        frame_data = get_image.get_frame()
+        
+        if frame_data is None:
+            print("⚠️ 无法获取帧")
+            continue
+        
+        frame_count += 1
+        if frame_count % 30 == 0:  # 每30帧显示一次状态
+            print(f"📊 已获取 {frame_count} 帧")
+        
+        # 显示图像
+        cv2.imshow('采集卡测试', frame_data[0])
+        
+        # 按 'q' 退出
+        if cv2.waitKey(1) & 0xFF == ord('q'):
+            break
+    
+    # 清理
+    get_image.release()
+    cv2.destroyAllWindows()
+    print("🔚 测试结束")
+
+if __name__ == "__main__":
+    test_camera()

test_capture_card.py

@@ -0,0 +1,796 @@
+"""
+采集卡截图测试类
+用于测试采集卡的分辨率和色差
+保持与实际代码相同的实现逻辑
+"""
+import time
+import threading
+import warnings
+import os
+import sys
+import io
+import cv2
+import numpy as np
+from PIL import Image
+from utils.logger import logger, throttle
+import logging
+
+# 抑制OpenCV的警告信息（兼容不同版本）
+os.environ['OPENCV_LOG_LEVEL'] = 'SILENT'
+os.environ['OPENCV_IO_ENABLE_OPENEXR'] = '0'
+
+try:
+    if hasattr(cv2, 'setLogLevel'):
+        if hasattr(cv2, 'LOG_LEVEL_SILENT'):
+            cv2.setLogLevel(cv2.LOG_LEVEL_SILENT)
+        elif hasattr(cv2, 'LOG_LEVEL_ERROR'):
+            cv2.setLogLevel(cv2.LOG_LEVEL_ERROR)
+        elif hasattr(cv2, 'utils'):
+            cv2.utils.setLogLevel(0)
+except Exception:
+    pass
+
+
+class CaptureCardTester:
+    """
+    采集卡测试类
+    使用与实际代码相同的实现逻辑
+    """
+    
+    def __init__(self, cam_index=0, width=1920, height=1080):
+        """
+        初始化采集卡测试器
+        :param cam_index: 采集卡索引
+        :param width: 期望宽度
+        :param height: 期望高度
+        """
+        logger.info(f"🔧 正在初始化采集卡测试器 {cam_index}...")
+        self.cap = None
+        self.frame = None
+        self.running = True
+        self.cam_index = cam_index
+        self.expected_width = width
+        self.expected_height = height
+        self.actual_width = None
+        self.actual_height = None
+        
+        # 尝试多种方式打开采集卡（与实际代码相同）
+        backends_to_try = [
+            (cam_index, cv2.CAP_DSHOW),
+            (cam_index, cv2.CAP_ANY),
+            (cam_index, None),  # 默认后端
+        ]
+        
+        # 重定向stderr来抑制OpenCV的错误输出
+        old_stderr = sys.stderr
+        suppressed_output = io.StringIO()
+        
+        try:
+            sys.stderr = suppressed_output
+            
+            for idx, backend in backends_to_try:
+                try:
+                    with warnings.catch_warnings():
+                        warnings.filterwarnings('ignore', category=UserWarning)
+                        if backend is not None:
+                            self.cap = cv2.VideoCapture(idx, backend)
+                        else:
+                            self.cap = cv2.VideoCapture(idx)
+                    
+                    if self.cap.isOpened():
+                        # 测试读取一帧
+                        ret, test_frame = self.cap.read()
+                        if ret and test_frame is not None:
+                            logger.info(f"✅ 采集卡 {cam_index} 打开成功")
+                            break
+                        else:
+                            self.cap.release()
+                            self.cap = None
+                except Exception as e:
+                    if self.cap:
+                        try:
+                            self.cap.release()
+                        except:
+                            pass
+                        self.cap = None
+                    continue
+        finally:
+            # 恢复stderr
+            sys.stderr = old_stderr
+        
+        if self.cap is None or not self.cap.isOpened():
+            logger.error(f"❌ 无法打开采集卡 {cam_index}")
+            logger.error("请检查：\n  1. 采集卡是否正确连接\n  2. 采集卡索引是否正确（尝试扫描采集卡）\n  3. 采集卡驱动是否安装\n  4. 采集卡是否被其他程序占用")
+            self.cap = None
+            return
+        
+        # 设置分辨率（与实际代码相同）
+        try:
+            self.cap.set(cv2.CAP_PROP_FRAME_WIDTH, width)
+            self.cap.set(cv2.CAP_PROP_FRAME_HEIGHT, height)
+            # 实际获取设置后的分辨率
+            self.actual_width = int(self.cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+            self.actual_height = int(self.cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+            logger.info(f"   分辨率设置: {width}x{height} -> 实际: {self.actual_width}x{self.actual_height}")
+        except Exception as e:
+            logger.warning(f"⚠️ 设置分辨率失败: {e}")
+        
+        # 启动更新线程（与实际代码相同）
+        threading.Thread(target=self.update, daemon=True).start()
+        
+        # 等待几帧确保采集卡正常工作
+        time.sleep(1.0)
+        logger.info(f"✅ 采集卡 {cam_index} 初始化完成")
+    
+    def update(self):
+        """持续更新帧（与实际代码相同）"""
+        while self.running and self.cap is not None:
+            try:
+                ret, frame = self.cap.read()
+                if ret and frame is not None:
+                    self.frame = frame
+                    # 限制读取频率，避免占满CPU
+                    time.sleep(0.008)
+                else:
+                    # 读取失败时不打印，避免刷屏
+                    time.sleep(0.02)
+            except Exception as e:
+                # 只在异常时打印错误
+                throttle(f"cap_read_err_{self.cam_index}", 2.0, logging.WARNING, f"⚠️ 采集卡 {self.cam_index} 读取异常: {e}")
+                time.sleep(0.1)  # 出错时短暂延迟
+    
+    def get_frame(self):
+        """
+        获取处理后的帧（与实际代码相同）
+        返回: [im_opencv, im_PIL] 或 None
+        """
+        if self.cap is None or self.frame is None:
+            return None
+        try:
+            im_opencv = cv2.cvtColor(self.frame, cv2.COLOR_BGR2RGB)
+            im_opencv = im_opencv[30:30+720, 0:1280]  # 裁剪尺寸（与实际代码相同）
+            im_PIL = Image.fromarray(im_opencv)
+            return [im_opencv, im_PIL]
+        except Exception as e:
+            throttle(f"img_proc_err_{self.cam_index}", 2.0, logging.WARNING, f"⚠️ 图像处理错误: {e}")
+            return None
+    
+    def get_raw_frame(self):
+        """
+        获取原始帧（未裁剪）
+        返回: numpy array 或 None
+        """
+        if self.cap is None or self.frame is None:
+            return None
+        return self.frame.copy()
+    
+    def test_resolution(self):
+        """
+        测试分辨率
+        返回分辨率信息字典
+        """
+        if self.cap is None:
+            logger.error("采集卡未初始化")
+            return None
+        
+        result = {
+            'expected': (self.expected_width, self.expected_height),
+            'actual_cap': (self.actual_width, self.actual_height),
+            'actual_frame': None,
+            'cropped_frame': None,
+            'match': False
+        }
+        
+        # 获取原始帧尺寸
+        raw_frame = self.get_raw_frame()
+        if raw_frame is not None:
+            h, w = raw_frame.shape[:2]
+            result['actual_frame'] = (w, h)
+        
+        # 获取裁剪后帧尺寸
+        processed = self.get_frame()
+        if processed is not None:
+            im_opencv = processed[0]
+            h, w = im_opencv.shape[:2]
+            result['cropped_frame'] = (w, h)
+        
+        # 检查分辨率是否匹配
+        if result['actual_cap'] is not None:
+            result['match'] = (result['actual_cap'][0] == self.expected_width and 
+                             result['actual_cap'][1] == self.expected_height)
+        
+        return result
+    
+    def test_color_difference(self, frame1=None, frame2=None):
+        """
+        测试色差
+        :param frame1: 第一帧（可选，不提供则使用当前帧）
+        :param frame2: 第二帧（可选，不提供则等待一帧后获取）
+        :return: 色差信息字典
+        """
+        if frame1 is None:
+            frame1 = self.get_frame()
+            if frame1 is None:
+                logger.error("无法获取第一帧")
+                return None
+            frame1 = frame1[0]  # 使用opencv格式
+        
+        if frame2 is None:
+            # 等待一小段时间获取新帧
+            time.sleep(0.1)
+            frame2 = self.get_frame()
+            if frame2 is None:
+                logger.error("无法获取第二帧")
+                return None
+            frame2 = frame2[0]  # 使用opencv格式
+        
+        # 确保两帧尺寸相同
+        if frame1.shape != frame2.shape:
+            logger.warning(f"两帧尺寸不同: {frame1.shape} vs {frame2.shape}")
+            # 调整尺寸
+            h, w = min(frame1.shape[0], frame2.shape[0]), min(frame1.shape[1], frame2.shape[1])
+            frame1 = frame1[:h, :w]
+            frame2 = frame2[:h, :w]
+        
+        # 计算色差
+        diff = cv2.absdiff(frame1, frame2)
+        diff_gray = cv2.cvtColor(diff, cv2.COLOR_RGB2GRAY)
+        
+        # 计算统计信息
+        mean_diff = np.mean(diff)
+        std_diff = np.std(diff)
+        max_diff = np.max(diff)
+        mean_diff_gray = np.mean(diff_gray)
+        
+        # 计算RGB各通道的平均色差
+        mean_diff_r = np.mean(diff[:, :, 0])
+        mean_diff_g = np.mean(diff[:, :, 1])
+        mean_diff_b = np.mean(diff[:, :, 2])
+        
+        # 计算PSNR（峰值信噪比）
+        mse = np.mean((frame1.astype(float) - frame2.astype(float)) ** 2)
+        if mse == 0:
+            psnr = float('inf')
+        else:
+            psnr = 20 * np.log10(255.0 / np.sqrt(mse))
+        
+        result = {
+            'mean_diff': float(mean_diff),
+            'std_diff': float(std_diff),
+            'max_diff': int(max_diff),
+            'mean_diff_gray': float(mean_diff_gray),
+            'mean_diff_r': float(mean_diff_r),
+            'mean_diff_g': float(mean_diff_g),
+            'mean_diff_b': float(mean_diff_b),
+            'psnr': float(psnr),
+            'mse': float(mse),
+            'diff_image': diff,
+            'diff_gray': diff_gray
+        }
+        
+        return result
+    
+    def test_color_stability(self, num_frames=10, interval=0.1):
+        """
+        测试颜色稳定性（连续多帧的色差）
+        :param num_frames: 测试帧数
+        :param interval: 帧间隔（秒）
+        :return: 稳定性统计信息
+        """
+        frames = []
+        logger.info(f"开始采集 {num_frames} 帧用于稳定性测试...")
+        
+        for i in range(num_frames):
+            frame = self.get_frame()
+            if frame is None:
+                logger.warning(f"无法获取第 {i+1} 帧")
+                continue
+            frames.append(frame[0])  # 使用opencv格式
+            if i < num_frames - 1:
+                time.sleep(interval)
+        
+        if len(frames) < 2:
+            logger.error("采集的帧数不足")
+            return None
+        
+        # 计算所有帧之间的平均色差
+        all_diffs = []
+        for i in range(len(frames) - 1):
+            diff_result = self.test_color_difference(frames[i], frames[i+1])
+            if diff_result:
+                all_diffs.append(diff_result['mean_diff'])
+        
+        if not all_diffs:
+            return None
+        
+        result = {
+            'num_frames': len(frames),
+            'avg_mean_diff': float(np.mean(all_diffs)),
+            'std_mean_diff': float(np.std(all_diffs)),
+            'min_mean_diff': float(np.min(all_diffs)),
+            'max_mean_diff': float(np.max(all_diffs)),
+            'all_diffs': [float(d) for d in all_diffs]
+        }
+        
+        return result
+    
+    def print_resolution_test(self):
+        """打印分辨率测试结果"""
+        print("\n" + "="*60)
+        print("分辨率测试结果")
+        print("="*60)
+        result = self.test_resolution()
+        if result is None:
+            print("❌ 测试失败：采集卡未初始化")
+            return
+        
+        print(f"期望分辨率: {result['expected'][0]} x {result['expected'][1]}")
+        if result['actual_cap']:
+            print(f"实际分辨率(采集卡): {result['actual_cap'][0]} x {result['actual_cap'][1]}")
+            match_str = "✅ 匹配" if result['match'] else "❌ 不匹配"
+            print(f"分辨率匹配: {match_str}")
+        
+        if result['actual_frame']:
+            print(f"实际分辨率(原始帧): {result['actual_frame'][0]} x {result['actual_frame'][1]}")
+        
+        if result['cropped_frame']:
+            print(f"裁剪后分辨率: {result['cropped_frame'][0]} x {result['cropped_frame'][1]}")
+        print("="*60 + "\n")
+    
+    def print_color_test(self):
+        """打印色差测试结果"""
+        print("\n" + "="*60)
+        print("色差测试结果（两帧对比）")
+        print("="*60)
+        result = self.test_color_difference()
+        if result is None:
+            print("❌ 测试失败：无法获取帧")
+            return
+        
+        print(f"平均色差: {result['mean_diff']:.2f}")
+        print(f"色差标准差: {result['std_diff']:.2f}")
+        print(f"最大色差: {result['max_diff']}")
+        print(f"灰度平均色差: {result['mean_diff_gray']:.2f}")
+        print(f"\nRGB通道平均色差:")
+        print(f"  R通道: {result['mean_diff_r']:.2f}")
+        print(f"  G通道: {result['mean_diff_g']:.2f}")
+        print(f"  B通道: {result['mean_diff_b']:.2f}")
+        print(f"\nPSNR (峰值信噪比): {result['psnr']:.2f} dB")
+        print(f"MSE (均方误差): {result['mse']:.2f}")
+        print("="*60 + "\n")
+    
+    def print_stability_test(self, num_frames=10):
+        """打印稳定性测试结果"""
+        print("\n" + "="*60)
+        print(f"颜色稳定性测试结果（{num_frames}帧）")
+        print("="*60)
+        result = self.test_color_stability(num_frames)
+        if result is None:
+            print("❌ 测试失败：无法获取足够的帧")
+            return
+        
+        print(f"测试帧数: {result['num_frames']}")
+        print(f"平均色差均值: {result['avg_mean_diff']:.2f}")
+        print(f"色差标准差: {result['std_mean_diff']:.2f}")
+        print(f"最小色差: {result['min_mean_diff']:.2f}")
+        print(f"最大色差: {result['max_mean_diff']:.2f}")
+        print(f"\n各帧间色差: {[f'{d:.2f}' for d in result['all_diffs']]}")
+        print("="*60 + "\n")
+    
+    def save_test_images(self, save_dir="test_output"):
+        """保存测试图像"""
+        os.makedirs(save_dir, exist_ok=True)
+        
+        # 保存原始帧
+        raw_frame = self.get_raw_frame()
+        if raw_frame is not None:
+            cv2.imwrite(os.path.join(save_dir, "raw_frame.jpg"), raw_frame)
+            logger.info(f"已保存原始帧: {save_dir}/raw_frame.jpg")
+        
+        # 保存处理后的帧
+        processed = self.get_frame()
+        if processed is not None:
+            cv2.imwrite(os.path.join(save_dir, "processed_frame.jpg"), cv2.cvtColor(processed[0], cv2.COLOR_RGB2BGR))
+            logger.info(f"已保存处理后帧: {save_dir}/processed_frame.jpg")
+        
+        # 保存色差图
+        color_diff = self.test_color_difference()
+        if color_diff is not None:
+            cv2.imwrite(os.path.join(save_dir, "color_diff.jpg"), cv2.cvtColor(color_diff['diff_image'], cv2.COLOR_RGB2BGR))
+            cv2.imwrite(os.path.join(save_dir, "color_diff_gray.jpg"), color_diff['diff_gray'])
+            logger.info(f"已保存色差图: {save_dir}/color_diff.jpg")
+            logger.info(f"已保存灰度色差图: {save_dir}/color_diff_gray.jpg")
+    
+    def release(self):
+        """释放资源（与实际代码相同）"""
+        self.running = False
+        time.sleep(0.2)
+        if self.cap is not None:
+            self.cap.release()
+        cv2.destroyAllWindows()
+        logger.info("🔚 采集卡已释放")
+    
+    def __del__(self):
+        """析构函数（与实际代码相同）"""
+        if hasattr(self, "cap") and self.cap is not None:
+            try:
+                self.release()
+            except:
+                pass
+
+
+class MultiCaptureCardTester:
+    """
+    多采集卡测试管理器
+    支持同时测试多张采集卡
+    """
+    
+    def __init__(self):
+        """初始化多采集卡测试管理器"""
+        self.testers = {}  # {cam_index: CaptureCardTester}
+        self.config = None
+    
+    def load_from_config(self):
+        """从配置文件加载采集卡"""
+        try:
+            from config import config_manager
+            config_manager.load_config()
+            self.config = config_manager.config
+            
+            groups = self.config.get('groups', [])
+            if not groups:
+                logger.warning("配置文件中没有找到配置组")
+                return []
+            
+            # 获取所有配置组中的采集卡
+            camera_configs = []
+            for group in groups:
+                cam_idx = group.get('camera_index')
+                cam_width = group.get('camera_width', 1920)
+                cam_height = group.get('camera_height', 1080)
+                name = group.get('name', f"配置组{groups.index(group)}")
+                
+                # 检查是否已存在相同索引的采集卡
+                if cam_idx not in [c['index'] for c in camera_configs]:
+                    camera_configs.append({
+                        'index': cam_idx,
+                        'width': cam_width,
+                        'height': cam_height,
+                        'name': name,
+                        'group': group
+                    })
+            
+            return camera_configs
+        except Exception as e:
+            logger.error(f"从配置加载失败: {e}")
+            return []
+    
+    def add_camera(self, cam_index, width=1920, height=1080, name=None):
+        """
+        添加一张采集卡到测试列表
+        :param cam_index: 采集卡索引
+        :param width: 宽度
+        :param height: 高度
+        :param name: 采集卡名称（可选）
+        """
+        if name is None:
+            name = f"采集卡{cam_index}"
+        
+        if cam_index in self.testers:
+            logger.warning(f"采集卡 {cam_index} 已存在，将重新初始化")
+            self.testers[cam_index].release()
+        
+        tester = CaptureCardTester(cam_index=cam_index, width=width, height=height)
+        if tester.cap is not None:
+            tester.name = name
+            self.testers[cam_index] = tester
+            return True
+        else:
+            logger.error(f"无法初始化采集卡 {cam_index}")
+            return False
+    
+    def initialize_from_config(self, use_all_groups=True):
+        """
+        从配置初始化所有采集卡
+        :param use_all_groups: 是否使用所有配置组，False则只使用活动配置组
+        """
+        camera_configs = self.load_from_config()
+        
+        if not camera_configs:
+            logger.warning("没有找到可用的采集卡配置")
+            return False
+        
+        # 筛选配置组
+        if not use_all_groups:
+            camera_configs = [c for c in camera_configs if c.get('group', {}).get('active', False)]
+            if not camera_configs:
+                logger.warning("没有活动的配置组")
+                return False
+        
+        logger.info(f"📷 找到 {len(camera_configs)} 张采集卡配置")
+        
+        success_count = 0
+        for config in camera_configs:
+            if self.add_camera(
+                cam_index=config['index'],
+                width=config['width'],
+                height=config['height'],
+                name=config['name']
+            ):
+                success_count += 1
+        
+        logger.info(f"✅ 成功初始化 {success_count}/{len(camera_configs)} 张采集卡")
+        return success_count > 0
+    
+    def test_all_resolution(self):
+        """测试所有采集卡的分辨率"""
+        print("\n" + "="*60)
+        print("所有采集卡分辨率测试")
+        print("="*60)
+        
+        results = {}
+        for cam_index, tester in self.testers.items():
+            name = getattr(tester, 'name', f"采集卡{cam_index}")
+            print(f"\n【{name} (索引: {cam_index})】")
+            result = tester.test_resolution()
+            results[cam_index] = result
+        
+        # 汇总结果
+        print("\n" + "-"*60)
+        print("分辨率测试汇总")
+        print("-"*60)
+        for cam_index, result in results.items():
+            if result is None:
+                continue
+            name = getattr(self.testers[cam_index], 'name', f"采集卡{cam_index}")
+            match_str = "✅" if result.get('match', False) else "❌"
+            print(f"{match_str} {name}: 期望{result['expected']} -> 实际{result.get('actual_cap', 'N/A')}")
+        
+        return results
+    
+    def test_all_color(self):
+        """测试所有采集卡的色差"""
+        print("\n" + "="*60)
+        print("所有采集卡色差测试")
+        print("="*60)
+        
+        results = {}
+        for cam_index, tester in self.testers.items():
+            name = getattr(tester, 'name', f"采集卡{cam_index}")
+            print(f"\n【{name} (索引: {cam_index})】")
+            result = tester.test_color_difference()
+            results[cam_index] = result
+        
+        # 汇总结果
+        print("\n" + "-"*60)
+        print("色差测试汇总")
+        print("-"*60)
+        for cam_index, result in results.items():
+            if result is None:
+                continue
+            name = getattr(self.testers[cam_index], 'name', f"采集卡{cam_index}")
+            print(f"{name}: 平均色差={result['mean_diff']:.2f}, PSNR={result['psnr']:.2f}dB")
+        
+        return results
+    
+    def test_all_stability(self, num_frames=10):
+        """测试所有采集卡的稳定性"""
+        print("\n" + "="*60)
+        print("所有采集卡稳定性测试")
+        print("="*60)
+        
+        results = {}
+        for cam_index, tester in self.testers.items():
+            name = getattr(tester, 'name', f"采集卡{cam_index}")
+            print(f"\n【{name} (索引: {cam_index})】")
+            result = tester.test_color_stability(num_frames=num_frames)
+            results[cam_index] = result
+        
+        # 汇总结果
+        print("\n" + "-"*60)
+        print("稳定性测试汇总")
+        print("-"*60)
+        for cam_index, result in results.items():
+            if result is None:
+                continue
+            name = getattr(self.testers[cam_index], 'name', f"采集卡{cam_index}")
+            print(f"{name}: 平均色差={result['avg_mean_diff']:.2f}±{result['std_mean_diff']:.2f}")
+        
+        return results
+    
+    def save_all_test_images(self, base_dir="test_output"):
+        """保存所有采集卡的测试图像"""
+        for cam_index, tester in self.testers.items():
+            name = getattr(tester, 'name', f"采集卡{cam_index}")
+            # 清理名称中的特殊字符，用于目录名
+            safe_name = "".join(c for c in name if c.isalnum() or c in (' ', '-', '_')).rstrip()
+            save_dir = os.path.join(base_dir, safe_name)
+            tester.save_test_images(save_dir=save_dir)
+    
+    def release_all(self):
+        """释放所有采集卡"""
+        for tester in self.testers.values():
+            try:
+                tester.release()
+            except:
+                pass
+        self.testers.clear()
+    
+    def __del__(self):
+        """析构函数"""
+        self.release_all()
+
+
+def scan_cameras(max_index=10):
+    """
+    扫描可用的采集卡
+    :param max_index: 最大扫描索引
+    :return: 可用采集卡索引列表
+    """
+    print("🔍 正在扫描采集卡...")
+    available = []
+    
+    old_stderr = sys.stderr
+    suppressed_output = io.StringIO()
+    
+    try:
+        sys.stderr = suppressed_output
+        
+        for i in range(max_index):
+            cap = None
+            try:
+                with warnings.catch_warnings():
+                    warnings.filterwarnings('ignore')
+                    cap = cv2.VideoCapture(i, cv2.CAP_DSHOW)
+                    if cap.isOpened():
+                        ret, frame = cap.read()
+                        if ret and frame is not None:
+                            available.append(i)
+                            print(f"  ✅ 找到采集卡: 索引 {i}")
+                if cap:
+                    cap.release()
+            except:
+                if cap:
+                    try:
+                        cap.release()
+                    except:
+                        pass
+    finally:
+        sys.stderr = old_stderr
+    
+    if not available:
+        print("  ❌ 未找到可用的采集卡")
+    else:
+        print(f"✅ 共找到 {len(available)} 张采集卡")
+    
+    return available
+
+
+def main():
+    """主测试函数"""
+    print("="*60)
+    print("采集卡截图测试工具（支持多采集卡）")
+    print("="*60)
+    
+    multi_tester = MultiCaptureCardTester()
+    
+    # 选择测试模式
+    print("\n选择测试模式:")
+    print("  1. 从配置文件加载（所有配置组）")
+    print("  2. 从配置文件加载（仅活动配置组）")
+    print("  3. 手动指定采集卡索引")
+    print("  4. 扫描采集卡")
+    print("  0. 退出")
+    
+    choice = input("\n请选择 (0-4): ").strip()
+    
+    if choice == "0":
+        print("👋 退出")
+        return
+    elif choice == "1":
+        # 从配置加载所有配置组
+        if not multi_tester.initialize_from_config(use_all_groups=True):
+            print("❌ 无法从配置初始化采集卡")
+            return
+    elif choice == "2":
+        # 从配置加载活动配置组
+        if not multi_tester.initialize_from_config(use_all_groups=False):
+            print("❌ 无法从配置初始化采集卡")
+            return
+    elif choice == "3":
+        # 手动指定
+        indices_input = input("请输入采集卡索引(用逗号分隔,如: 0,1,2): ").strip()
+        try:
+            indices = [int(x.strip()) for x in indices_input.split(',')]
+            width_input = input("请输入宽度 (默认1920): ").strip()
+            height_input = input("请输入高度 (默认1080): ").strip()
+            width = int(width_input) if width_input else 1920
+            height = int(height_input) if height_input else 1080
+            
+            success_count = 0
+            for idx in indices:
+                if multi_tester.add_camera(idx, width=width, height=height):
+                    success_count += 1
+            
+            if success_count == 0:
+                print("❌ 无法初始化任何采集卡")
+                return
+        except ValueError:
+            print("❌ 输入格式错误")
+            return
+    elif choice == "4":
+        # 扫描采集卡
+        available = scan_cameras()
+        if not available:
+            return
+        
+        indices_input = input(f"请输入要测试的采集卡索引(用逗号分隔,可用: {available}): ").strip()
+        try:
+            indices = [int(x.strip()) for x in indices_input.split(',')]
+            # 验证索引有效性
+            indices = [i for i in indices if i in available]
+            if not indices:
+                print("❌ 没有有效的采集卡索引")
+                return
+            
+            width_input = input("请输入宽度 (默认1920): ").strip()
+            height_input = input("请输入高度 (默认1080): ").strip()
+            width = int(width_input) if width_input else 1920
+            height = int(height_input) if height_input else 1080
+            
+            success_count = 0
+            for idx in indices:
+                if multi_tester.add_camera(idx, width=width, height=height):
+                    success_count += 1
+            
+            if success_count == 0:
+                print("❌ 无法初始化任何采集卡")
+                return
+        except ValueError:
+            print("❌ 输入格式错误")
+            return
+    else:
+        print("❌ 无效选择")
+        return
+    
+    if not multi_tester.testers:
+        print("❌ 没有可用的采集卡")
+        return
+    
+    try:
+        # 等待采集卡稳定
+        print("\n等待采集卡稳定...")
+        time.sleep(1.0)
+        
+        # 测试分辨率
+        multi_tester.test_all_resolution()
+        
+        # 等待一下确保采集卡稳定
+        time.sleep(0.5)
+        
+        # 测试色差
+        multi_tester.test_all_color()
+        
+        # 测试稳定性
+        multi_tester.test_all_stability(num_frames=10)
+        
+        # 保存测试图像
+        multi_tester.save_all_test_images()
+        
+        print("\n✅ 所有测试完成！")
+        print("按 Enter 键退出...")
+        input()
+        
+    except KeyboardInterrupt:
+        print("\n\n用户中断测试")
+    except Exception as e:
+        logger.error(f"测试过程中发生错误: {e}")
+        import traceback
+        traceback.print_exc()
+    finally:
+        multi_tester.release_all()
+
+
+if __name__ == "__main__":
+    main()
+

utils/device_scanner.py

@@ -0,0 +1,197 @@
+"""
+设备扫描工具 - 通过唯一标识符而非索引来识别设备
+解决重启后设备序号混乱的问题
+"""
+import cv2
+import serial.tools.list_ports
+import warnings
+import sys
+import io
+from typing import List, Dict, Optional
+
+def get_camera_info(cam_index: int) -> Optional[Dict]:
+    """
+    获取采集卡的详细信息（包括设备名称）
+    :param cam_index: 采集卡索引
+    :return: 包含设备信息的字典，如果失败返回None
+    """
+    old_stderr = sys.stderr
+    suppressed_output = io.StringIO()
+    
+    try:
+        sys.stderr = suppressed_output
+        
+        with warnings.catch_warnings():
+            warnings.filterwarnings('ignore')
+            
+            # 尝试多种后端
+            backends = [
+                (cam_index, cv2.CAP_DSHOW),
+                (cam_index, cv2.CAP_ANY),
+                (cam_index, None),
+            ]
+            
+            cap = None
+            for idx, backend in backends:
+                try:
+                    if backend is not None:
+                        cap = cv2.VideoCapture(idx, backend)
+                    else:
+                        cap = cv2.VideoCapture(idx)
+                    
+                    if cap.isOpened():
+                        ret, frame = cap.read()
+                        if ret and frame is not None:
+                            # 尝试获取设备名称（不同后端可能支持不同）
+                            device_name = None
+                            backend_name = None
+                            
+                            if backend == cv2.CAP_DSHOW:
+                                backend_name = "DirectShow"
+                                # DirectShow可能支持获取设备名称
+                                try:
+                                    # 某些情况下可以通过属性获取
+                                    pass  # OpenCV限制，无法直接获取设备名称
+                                except:
+                                    pass
+                            elif backend == cv2.CAP_ANY:
+                                backend_name = "Any"
+                            else:
+                                backend_name = "Default"
+                            
+                            # 获取分辨率信息作为辅助标识
+                            width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+                            height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+                            
+                            cap.release()
+                            
+                            return {
+                                'index': cam_index,
+                                'backend': backend_name,
+                                'width': width,
+                                'height': height,
+                                'name': f"采集卡-{cam_index}",
+                                'available': True
+                            }
+                        else:
+                            if cap:
+                                cap.release()
+                                cap = None
+                except Exception:
+                    if cap:
+                        try:
+                            cap.release()
+                        except:
+                            pass
+                        cap = None
+                    continue
+            
+            return None
+    finally:
+        sys.stderr = old_stderr
+
+def scan_cameras_with_info(max_index: int = 10) -> List[Dict]:
+    """
+    扫描所有可用采集卡，返回详细信息列表
+    :param max_index: 最大扫描索引
+    :return: 采集卡信息列表
+    """
+    cameras = []
+    for i in range(max_index + 1):
+        info = get_camera_info(i)
+        if info:
+            cameras.append(info)
+    return cameras
+
+def get_serial_port_info(port_name: str) -> Optional[Dict]:
+    """
+    获取串口的详细信息（包括硬件ID和描述）
+    :param port_name: 串口名称（如 "COM3"）
+    :return: 包含串口信息的字典
+    """
+    try:
+        ports = serial.tools.list_ports.comports()
+        for port in ports:
+            if port.device == port_name:
+                return {
+                    'device': port.device,
+                    'description': port.description,
+                    'hwid': port.hwid,  # 硬件ID（唯一标识）
+                    'vid': port.vid if hasattr(port, 'vid') else None,
+                    'pid': port.pid if hasattr(port, 'pid') else None,
+                    'serial_number': port.serial_number if hasattr(port, 'serial_number') else None,
+                    'manufacturer': port.manufacturer if hasattr(port, 'manufacturer') else None,
+                    'name': port.description or port.device
+                }
+    except Exception as e:
+        print(f"⚠️ 获取串口信息失败: {e}")
+    
+    return None
+
+def scan_serial_ports_with_info() -> List[Dict]:
+    """
+    扫描所有可用串口，返回详细信息列表
+    :return: 串口信息列表
+    """
+    ports_info = []
+    try:
+        ports = serial.tools.list_ports.comports()
+        for port in ports:
+            info = {
+                'device': port.device,
+                'description': port.description,
+                'hwid': port.hwid,
+                'vid': port.vid if hasattr(port, 'vid') else None,
+                'pid': port.pid if hasattr(port, 'pid') else None,
+                'serial_number': port.serial_number if hasattr(port, 'serial_number') else None,
+                'manufacturer': port.manufacturer if hasattr(port, 'manufacturer') else None,
+                'name': port.description or port.device
+            }
+            ports_info.append(info)
+    except Exception as e:
+        print(f"⚠️ 扫描串口失败: {e}")
+    
+    # 按设备名排序
+    ports_info.sort(key=lambda x: int(x['device'].replace('COM', '')) if x['device'].replace('COM', '').isdigit() else 999)
+    return ports_info
+
+def find_camera_by_index(cameras: List[Dict], index: int) -> Optional[Dict]:
+    """通过索引查找采集卡"""
+    for cam in cameras:
+        if cam['index'] == index:
+            return cam
+    return None
+
+def find_serial_by_hwid(ports: List[Dict], hwid: str) -> Optional[Dict]:
+    """通过硬件ID查找串口（最可靠）"""
+    for port in ports:
+        if port['hwid'] == hwid:
+            return port
+    return None
+
+def find_serial_by_device(ports: List[Dict], device: str) -> Optional[Dict]:
+    """通过设备名查找串口（兼容旧配置）"""
+    for port in ports:
+        if port['device'] == device:
+            return port
+    return None
+
+if __name__ == "__main__":
+    # 测试代码
+    print("=" * 60)
+    print("采集卡扫描:")
+    print("=" * 60)
+    cameras = scan_cameras_with_info()
+    for cam in cameras:
+        print(f"  索引 {cam['index']}: {cam['name']} ({cam['width']}x{cam['height']})")
+    
+    print("\n" + "=" * 60)
+    print("串口扫描:")
+    print("=" * 60)
+    ports = scan_serial_ports_with_info()
+    for port in ports:
+        print(f"  {port['device']}: {port['name']}")
+        print(f"    硬件ID: {port['hwid']}")
+        if port['vid'] and port['pid']:
+            print(f"    VID/PID: {port['vid']:04X}/{port['pid']:04X}")
+

utils/get_image.py

@@ -2,6 +2,8 @@ import time
 
 from PIL import Image
 import cv2
+from utils.logger import logger, throttle
+import logging
 # class GetImage:
 #     def __init__(self, cam_index=0, width=1920, height=1080):
 #         self.cap = cv2.VideoCapture(cam_index,cv2.CAP_DSHOW)
@@ -41,33 +43,137 @@ import cv2
 #     cv2.destroyAllWindows()
 
 import threading
+import warnings
+
+# 抑制OpenCV的警告信息（兼容不同版本）
+import os
+import sys
+import io
+os.environ['OPENCV_LOG_LEVEL'] = 'SILENT'
+os.environ['OPENCV_IO_ENABLE_OPENEXR'] = '0'
+
+try:
+    if hasattr(cv2, 'setLogLevel'):
+        if hasattr(cv2, 'LOG_LEVEL_SILENT'):
+            cv2.setLogLevel(cv2.LOG_LEVEL_SILENT)
+        elif hasattr(cv2, 'LOG_LEVEL_ERROR'):
+            cv2.setLogLevel(cv2.LOG_LEVEL_ERROR)
+        elif hasattr(cv2, 'utils'):
+            cv2.utils.setLogLevel(0)
+except Exception:
+    pass
 
 class GetImage:
     def __init__(self, cam_index=0, width=1920, height=1080):
-        self.cap = cv2.VideoCapture(cam_index, cv2.CAP_DSHOW)
-        self.cap.set(cv2.CAP_PROP_FRAME_WIDTH, width)
-        self.cap.set(cv2.CAP_PROP_FRAME_HEIGHT, height)
+        logger.info(f"🔧 正在初始化采集卡 {cam_index}...")
+        self.cap = None
         self.frame = None
         self.running = True
+        self.cam_index = cam_index
+        
+        # 尝试多种方式打开采集卡
+        backends_to_try = [
+            (cam_index, cv2.CAP_DSHOW),
+            (cam_index, cv2.CAP_ANY),
+            (cam_index, None),  # 默认后端
+        ]
+        
+        # 重定向stderr来抑制OpenCV的错误输出
+        old_stderr = sys.stderr
+        suppressed_output = io.StringIO()
+        
+        try:
+            sys.stderr = suppressed_output
+            
+            for idx, backend in backends_to_try:
+                try:
+                    with warnings.catch_warnings():
+                        warnings.filterwarnings('ignore', category=UserWarning)
+                        if backend is not None:
+                            self.cap = cv2.VideoCapture(idx, backend)
+                        else:
+                            self.cap = cv2.VideoCapture(idx)
+                    
+                    if self.cap.isOpened():
+                        # 测试读取一帧
+                        ret, test_frame = self.cap.read()
+                        if ret and test_frame is not None:
+                            logger.info(f"✅ 采集卡 {cam_index} 打开成功")
+                            break
+                        else:
+                            self.cap.release()
+                            self.cap = None
+                except Exception as e:
+                    if self.cap:
+                        try:
+                            self.cap.release()
+                        except:
+                            pass
+                        self.cap = None
+                    continue
+        finally:
+            # 恢复stderr
+            sys.stderr = old_stderr
+        
+        if self.cap is None or not self.cap.isOpened():
+            logger.error(f"❌ 无法打开采集卡 {cam_index}")
+            logger.error("请检查：\n  1. 采集卡是否正确连接\n  2. 采集卡索引是否正确（尝试扫描采集卡）\n  3. 采集卡驱动是否安装\n  4. 采集卡是否被其他程序占用")
+            self.cap = None
+            return
+        
+        # 设置分辨率
+        try:
+            self.cap.set(cv2.CAP_PROP_FRAME_WIDTH, width)
+            self.cap.set(cv2.CAP_PROP_FRAME_HEIGHT, height)
+            # 实际获取设置后的分辨率
+            actual_width = int(self.cap.get(cv2.CAP_PROP_FRAME_WIDTH))
+            actual_height = int(self.cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
+            logger.info(f"   分辨率设置: {width}x{height} -> 实际: {actual_width}x{actual_height}")
+        except Exception as e:
+            logger.warning(f"⚠️ 设置分辨率失败: {e}")
+        
+        # 启动更新线程
         threading.Thread(target=self.update, daemon=True).start()
         
+        # 等待几帧确保采集卡正常工作
+        import time
+        time.sleep(1.0)
+        logger.info(f"✅ 采集卡 {cam_index} 初始化完成")
+
     def update(self):
-        while self.running:
+        while self.running and self.cap is not None:
+            try:
                 ret, frame = self.cap.read()
-            if ret:
+                if ret and frame is not None:
                     self.frame = frame
+                    # 限制读取频率，避免占满CPU
+                    time.sleep(0.008)
+                else:
+                    # 读取失败时不打印，避免刷屏
+                    time.sleep(0.02)
+            except Exception as e:
+                # 只在异常时打印错误
+                throttle(f"cap_read_err_{self.cam_index}", 2.0, logging.WARNING, f"⚠️ 采集卡 {self.cam_index} 读取异常: {e}")
+                time.sleep(0.1)  # 出错时短暂延迟
 
     def get_frame(self):
-        if self.frame is None:
+        if self.cap is None or self.frame is None:
             return None
+        try:
             im_opencv = cv2.cvtColor(self.frame, cv2.COLOR_BGR2RGB)
             im_opencv = im_opencv[30:30+720, 0:1280]
             im_PIL = Image.fromarray(im_opencv)
             return [im_opencv, im_PIL]
+        except Exception as e:
+            throttle(f"img_proc_err_{self.cam_index}", 2.0, logging.WARNING, f"⚠️ 图像处理错误: {e}")
+            return None
 
     def release(self):
         self.running = False
         time.sleep(0.2)
+        if self.cap is not None:
             self.cap.release()
         cv2.destroyAllWindows()
-get_image = GetImage()
+
+# get_image 将在main.py中初始化
+get_image = None

utils/logger.py

@@ -0,0 +1,26 @@
+import logging
+import sys
+import time
+from typing import Dict
+
+# 基础logger配置（控制台输出）
+_logger = logging.getLogger("huojv")
+if not _logger.handlers:
+    _logger.setLevel(logging.INFO)
+    handler = logging.StreamHandler(sys.stdout)
+    fmt = logging.Formatter(fmt='[%(asctime)s] %(levelname)s %(message)s', datefmt='%H:%M:%S')
+    handler.setFormatter(fmt)
+    _logger.addHandler(handler)
+
+# 简单的节流打印：同一个key在interval秒内只打印一次
+_last_log_times: Dict[str, float] = {}
+
+def throttle(key: str, interval_sec: float, level: int, msg: str):
+    now = time.time()
+    last = _last_log_times.get(key, 0.0)
+    if now - last >= interval_sec:
+        _last_log_times[key] = now
+        _logger.log(level, msg)
+
+# 对外暴露
+logger = _logger

utils/mouse.py

@@ -1,17 +1,48 @@
 import random
 import time
 import ch9329Comm
-import time
-import random
 import serial
-serial.ser = serial.Serial('COM6', 9600)  # 开启串口
-mouse = ch9329Comm.mouse.DataComm(1920, 1080)
+
+# 全局变量，由main.py初始化
+serial.ser = None
+mouse = None
+
+def init_mouse_keyboard(config_group):
+    """初始化鼠标和串口"""
+    global serial, mouse
+    from utils.logger import logger
+    
+    # 初始化串口
+    try:
+        logger.info(f"🔧 正在打开串口: {config_group['serial_port']} @ {config_group['serial_baudrate']}")
+        serial.ser = serial.Serial(
+            config_group['serial_port'], 
+            config_group['serial_baudrate'],
+            timeout=1
+        )
+        logger.info(f"✅ 串口已打开: {config_group['serial_port']} @ {config_group['serial_baudrate']}")
+    except Exception as e:
+        logger.error(f"❌ 串口打开失败: {e}")
+        raise
+    
+    # 初始化鼠标
+    try:
+        logger.info(f"🔧 正在初始化鼠标: {config_group['camera_width']}x{config_group['camera_height']}")
+        mouse = ch9329Comm.mouse.DataComm(
+            config_group['camera_width'], 
+            config_group['camera_height']
+        )
+        logger.info(f"✅ 鼠标已初始化: {config_group['camera_width']}x{config_group['camera_height']}")
+    except Exception as e:
+        logger.error(f"❌ 鼠标初始化失败: {e}")
+        raise
 
 def bezier_point(t, p0, p1, p2, p3):
     """计算三次贝塞尔曲线上的点"""
     x = (1-t)**3 * p0[0] + 3*(1-t)**2*t*p1[0] + 3*(1-t)*t**2*p2[0] + t**3*p3[0]
     y = (1-t)**3 * p0[1] + 3*(1-t)**2*t*p1[1] + 3*(1-t)*t**2*p2[1] + t**3*p3[1]
     return (x, y)
+
 def move_mouse_bezier(mouse, start, end, duration=1, steps=120):
     """
     用贝塞尔曲线模拟鼠标移动（安全版）
@@ -45,13 +76,14 @@ class Mouse_guiji():
     def __init__(self):
         self.point=(0,0)
 
-    def send_data_absolute(self, x, y,may=0):
+    def send_data_absolute(self, x, y, may=0):
         move_mouse_bezier(mouse, self.point, (x,y), duration=1, steps=120)
-        if may == 1:#点击左
+        if may == 1:  # 点击左
             mouse.click()
         elif may == 2:
-            mouse.click1()#点击右
+            mouse.click1()  # 点击右
         self.point=(x,y)
 
-mouse_gui = Mouse_guiji()
+# mouse_gui 将在main.py中初始化
+mouse_gui = None
 

utils/shizi.py

@@ -45,7 +45,7 @@ def tiaozhan(image):
 
 
 def tuichu(image):
-    image = image[24:58, 569:669]
+    image = image[36:58, 560:669]
     image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
     # 将裁剪后的图像编码成二进制格式
     _, img_encoded = cv2.imencode('.png', image)
@@ -68,7 +68,7 @@ def tuwai(image):
     img_bytes = img_encoded.tobytes()
     result = ocr.classification(img_bytes)
     print(result)
-    if result == "tap" or result=='tqp' or result=='top':
+    if result == "tap" or result=='tqp' or result=='top' or result=='tp':
         return True
     else:
         return False

yolo_test.py

@@ -1,42 +1,344 @@
 import cv2
-from utils.get_image import get_image
+from utils.get_image import GetImage
 from ultralytics import YOLO
+from config import config_manager
+from utils.logger import logger
+import os
+import numpy as np
 
-model = YOLO(r"best0.pt").to('cuda')
+# 检查模型文件是否存在
+model_path = r"best0.pt"
+if not os.path.exists(model_path):
+    print(f"❌ 模型文件不存在: {model_path}")
+    exit(1)
 
-def yolo_shibie(im_PIL, detections):
+# 加载YOLO模型
+try:
+    model = YOLO(model_path).to('cuda')
+    print(f"✅ 模型加载成功: {model_path}")
+except Exception as e:
+    print(f"❌ 模型加载失败: {e}")
+    exit(1)
+
+def enhance_sharpness(image, strength=1.5):
+    """
+    增强图像锐度
+    :param image: 输入图像（BGR格式）
+    :param strength: 锐化强度（1.0-3.0，默认1.5）
+    :return: 锐化后的图像
+    """
+    # 创建锐化核
+    kernel = np.array([[-1, -1, -1],
+                       [-1,  9*strength, -1],
+                       [-1, -1, -1]]) / (9*strength - 8)
+    sharpened = cv2.filter2D(image, -1, kernel)
+    return sharpened
+
+
+def enhance_contrast(image, alpha=1.2, beta=10):
+    """
+    增强对比度和亮度
+    :param image: 输入图像
+    :param alpha: 对比度控制（1.0-3.0，默认1.2）
+    :param beta: 亮度控制（-100到100，默认10）
+    :return: 增强后的图像
+    """
+    return cv2.convertScaleAbs(image, alpha=alpha, beta=beta)
+
+
+def denoise_image(image, method='bilateral'):
+    """
+    去噪处理
+    :param image: 输入图像
+    :param method: 去噪方法 ('bilateral', 'gaussian', 'fastNlMeans')
+    :return: 去噪后的图像
+    """
+    if method == 'bilateral':
+        # 双边滤波，保留边缘的同时去噪
+        return cv2.bilateralFilter(image, 9, 75, 75)
+    elif method == 'gaussian':
+        # 高斯模糊去噪
+        return cv2.GaussianBlur(image, (5, 5), 0)
+    elif method == 'fastNlMeans':
+        # 非局部均值去噪（效果最好但较慢）
+        return cv2.fastNlMeansDenoisingColored(image, None, 10, 10, 7, 21)
+    return image
+
+
+def apply_enhancements(image, sharpness=True, contrast=True, denoise=True, 
+                       sharp_strength=1.5, contrast_alpha=1.2, contrast_beta=10, 
+                       denoise_method='bilateral'):
+    """
+    应用所有图像增强
+    :param image: 输入图像（BGR格式）
+    :param sharpness: 是否锐化
+    :param contrast: 是否增强对比度
+    :param denoise: 是否去噪
+    :param sharp_strength: 锐化强度
+    :param contrast_alpha: 对比度系数
+    :param contrast_beta: 亮度调整
+    :param denoise_method: 去噪方法
+    :return: 增强后的图像
+    """
+    enhanced = image.copy()
+    
+    if denoise:
+        enhanced = denoise_image(enhanced, denoise_method)
+    
+    if contrast:
+        enhanced = enhance_contrast(enhanced, contrast_alpha, contrast_beta)
+    
+    if sharpness:
+        enhanced = enhance_sharpness(enhanced, sharp_strength)
+    
+    return enhanced
+
+
+def set_camera_properties(cap, brightness=None, contrast=None, saturation=None, 
+                          sharpness=None, gain=None, exposure=None):
+    """
+    设置采集卡硬件参数
+    :param cap: VideoCapture对象
+    :param brightness: 亮度 (0-100)
+    :param contrast: 对比度 (0-100)
+    :param saturation: 饱和度 (0-100)
+    :param sharpness: 锐度 (0-100)
+    :param gain: 增益 (0-100)
+    :param exposure: 曝光 (通常为负值，如-6)
+    """
+    props = {
+        cv2.CAP_PROP_BRIGHTNESS: brightness,
+        cv2.CAP_PROP_CONTRAST: contrast,
+        cv2.CAP_PROP_SATURATION: saturation,
+        cv2.CAP_PROP_SHARPNESS: sharpness,
+        cv2.CAP_PROP_GAIN: gain,
+        cv2.CAP_PROP_EXPOSURE: exposure,
+    }
+    
+    for prop, value in props.items():
+        if value is not None:
+            try:
+                cap.set(prop, value)
+                actual = cap.get(prop)
+                logger.info(f"  设置 {prop.name if hasattr(prop, 'name') else prop}: {value} -> 实际: {actual:.2f}")
+            except Exception as e:
+                logger.warning(f"  ⚠️ 设置参数 {prop} 失败: {e}")
+
+
+def yolo_shibie(im_PIL, detections, model, enhance_enabled=False, enhance_params=None):
+    """
+    YOLO识别函数
+    :param im_PIL: PIL图像对象
+    :param detections: 检测结果字典
+    :param model: YOLO模型
+    :param enhance_enabled: 是否启用图像增强
+    :param enhance_params: 图像增强参数
+    :return: 更新后的detections字典，如果用户退出则返回None
+    """
+    if im_PIL is None:
+        return detections
+    
+    try:
         results = model(im_PIL)
         result = results[0]
 
-    # ✅ 获取绘制好框的图像
-    frame_with_boxes = result.plot()
+        # ✅ 获取绘制好框的图像（RGB格式）
+        frame_with_boxes_rgb = result.plot()
         
-    # ✅ 用 OpenCV 动态显示
-    cv2.imshow("YOLO实时检测", frame_with_boxes)
+        # ✅ 转换为BGR格式用于OpenCV显示
+        frame_with_boxes_bgr = cv2.cvtColor(frame_with_boxes_rgb, cv2.COLOR_RGB2BGR)
 
-    # ESC 或 Q 键退出
-    if cv2.waitKey(1) & 0xFF in [27, ord('q')]:
-        return None
+        # 应用图像增强（如果启用）
+        display_frame = frame_with_boxes_bgr.copy()
+        if enhance_enabled and enhance_params:
+            try:
+                display_frame = apply_enhancements(display_frame, **enhance_params)
+            except Exception as e:
+                print(f"⚠️ 图像增强失败: {e}")
+        
+        # 显示YOLO检测结果
+        cv2.imshow("YOLO Real-time Detection", display_frame)
 
         # ✅ 提取检测信息
+        if result.boxes is not None and len(result.boxes.xyxy) > 0:
+            # 用于存储多个候选npc4（如果检测到多个）
+            npc4_candidates = []
+            
             for i in range(len(result.boxes.xyxy)):
-        left, top, right, bottom = result.boxes.xyxy[i]
+                try:
+                    left = float(result.boxes.xyxy[i][0])
+                    top = float(result.boxes.xyxy[i][1])
+                    right = float(result.boxes.xyxy[i][2])
+                    bottom = float(result.boxes.xyxy[i][3])
                     cls_id = int(result.boxes.cls[i])
                     label = result.names[cls_id]
                     
-        if label in ['center', 'next', 'npc1', 'npc2', 'npc3', 'npc4', 'boss', 'zhaozi']:
+                    # 获取置信度（如果可用）
+                    confidence = float(result.boxes.conf[i]) if hasattr(result.boxes, 'conf') and len(result.boxes.conf) > i else 1.0
+
+                    # npc1-npc4 使用底部位置（与main.py保持一致）
+                    if label in ['npc1', 'npc2', 'npc3', 'npc4']:
+                        player_x = int(left + (right - left) / 2)
+                        player_y = int(bottom) + 30  # 使用底部位置，与main.py保持一致
+                        position = [player_x, player_y]
+                        
+                        # 特殊处理npc4：如果检测到多个，收集所有候选
+                        if label == 'npc4':
+                            npc4_candidates.append({
+                                'position': position,
+                                'confidence': confidence,
+                                'box': [left, top, right, bottom],
+                                'area': (right - left) * (bottom - top)  # 检测框面积
+                            })
+                        else:
+                            # npc1-npc3直接赋值（如果已经有值，保留置信度更高的）
+                            if detections[label] is None or (hasattr(result.boxes, 'conf') and 
+                                                              confidence > 0.5):
+                                detections[label] = position
+                    
+                    # 其他目标使用中心点
+                    elif label in ['center', 'next', 'boss', 'zhaozi']:
                         player_x = int(left + (right - left) / 2) + 3
                         player_y = int(top + (bottom - top) / 2) + 40
                         detections[label] = [player_x, player_y]
+                    
+                    # 道具和怪物可以多个
                     elif label in ['daojv', 'gw']:
                         player_x = int(left + (right - left) / 2) + 3
                         player_y = int(top + (bottom - top) / 2) + 40
+                        # 确保列表存在
+                        if label not in detections:
+                            detections[label] = []
                         detections[label].append([player_x, player_y])
                         
+                except Exception as e:
+                    print(f"⚠️ 处理检测框时出错: {e}")
+                    continue
+            
+            # 处理npc4：如果检测到多个，选择最合适的
+            if npc4_candidates:
+                # 按置信度排序，选择置信度最高的
+                npc4_candidates.sort(key=lambda x: x['confidence'], reverse=True)
+                
+                # 选择最佳候选（置信度最高且面积合理）
+                best_npc4 = None
+                for candidate in npc4_candidates:
+                    # 置信度阈值：至少0.3（可根据实际情况调整）
+                    if candidate['confidence'] >= 0.3:
+                        # 检查检测框面积是否合理（避免过小的误检）
+                        area = candidate['area']
+                        if area > 100:  # 最小面积阈值
+                            best_npc4 = candidate
+                            break
+                
+                if best_npc4:
+                    detections['npc4'] = best_npc4['position']
+                    # 可选：输出调试信息
+                    # print(f"✅ 检测到npc4: 位置={best_npc4['position']}, 置信度={best_npc4['confidence']:.2f}")
+                elif len(npc4_candidates) == 1:
+                    # 如果只有一个候选，即使置信度较低也使用
+                    detections['npc4'] = npc4_candidates[0]['position']
+
+    except Exception as e:
+        print(f"⚠️ YOLO检测出错: {e}")
+    
     return detections
 
 
-while True:
+def main():
+    """主函数"""
+    print("="*60)
+    print("YOLO实时检测测试")
+    print("="*60)
+    
+    # 从配置加载采集卡设置
+    active_group = config_manager.get_active_group()
+    
+    if active_group is None:
+        print("⚠️ 没有活动的配置组，使用默认设置")
+        print("提示: 可以运行 python gui_config.py 设置配置")
+        cam_index = 0
+        width = 1920
+        height = 1080
+    else:
+        print(f"📋 使用配置组: {active_group['name']}")
+        cam_index = active_group['camera_index']
+        width = active_group['camera_width']
+        height = active_group['camera_height']
+    
+    print(f"   采集卡索引: {cam_index}")
+    print(f"   分辨率: {width}x{height}")
+    print()
+    
+    # 初始化采集卡
+    print("🔧 正在初始化采集卡...")
+    get_image = GetImage(
+        cam_index=cam_index,
+        width=width,
+        height=height
+    )
+    
+    if get_image.cap is None:
+        print("❌ 采集卡初始化失败")
+        print("请检查：")
+        print("1. 采集卡是否正确连接")
+        print("2. 采集卡索引是否正确")
+        print("3. 采集卡驱动是否安装")
+        return
+    
+    # 设置采集卡硬件参数以提高清晰度（可选）
+    print("\n🔧 设置采集卡参数以提高清晰度...")
+    print("提示: 可以根据实际情况调整这些参数")
+    set_camera_properties(
+        get_image.cap,
+        brightness=50,      # 亮度 (0-100)
+        contrast=50,         # 对比度 (0-100)
+        saturation=55,      # 饱和度 (0-100)
+        sharpness=60,       # 锐度 (0-100，提高清晰度)
+        gain=None,          # 增益 (根据实际情况调整)
+        exposure=None       # 曝光 (根据实际情况调整，通常为负值)
+    )
+    
+    print("✅ 采集卡初始化成功")
+    print("\n快捷键:")
+    print("  'q' 或 ESC - 退出")
+    print("  'e' - 切换图像增强")
+    print("  '1'/'2' - 调整锐化强度 (+/-0.1)")
+    print("  '3'/'4' - 调整对比度 (+/-0.1)")
+    print()
+    
+    try:
+        frame_count = 0
+        enhance_enabled = False  # 默认关闭图像增强
+        
+        # 图像增强参数
+        enhance_params = {
+            'sharpness': True,
+            'contrast': True,
+            'denoise': True,
+            'sharp_strength': 1.5,
+            'contrast_alpha': 1.2,
+            'contrast_beta': 10,
+            'denoise_method': 'bilateral'
+        }
+        
+        while True:
+            # 获取帧
+            frame_data = get_image.get_frame()
+            
+            if frame_data is None:
+                print("⚠️ 无法获取帧，跳过...")
+                continue
+            
+            # frame_data 是 [im_opencv_rgb, im_PIL] 格式
+            # im_opencv_rgb 已经是RGB格式（经过BGR2RGB转换）
+            im_opencv_rgb, im_PIL = frame_data
+            
+            if im_PIL is None:
+                print("⚠️ PIL图像为空，跳过...")
+                continue
+            
+            # 初始化检测结果字典
             detections = {
                 'center': None, 'next': None,
                 'npc1': None, 'npc2': None, 'npc3': None, 'npc4': None,
@@ -44,12 +346,52 @@ while True:
                 'daojv': [], 'gw': []
             }
             
-    im_opencv = get_image.get_frame()  # [RGB, PIL]
-    detections = yolo_shibie(im_opencv[1], detections)
+            # 执行YOLO检测
+            detections = yolo_shibie(im_PIL, detections, model, enhance_enabled, enhance_params)
             
-    if detections is None:  # 用户退出
+            # 检查按键
+            key = cv2.waitKey(1) & 0xFF
+            if key in [27, ord('q'), ord('Q')]:
+                print("\n用户退出")
                 break
+            elif key == ord('e') or key == ord('E'):
+                enhance_enabled = not enhance_enabled
+                status = "开启" if enhance_enabled else "关闭"
+                print(f"图像增强: {status} (锐化={enhance_params['sharp_strength']:.1f}, "
+                      f"对比度={enhance_params['contrast_alpha']:.1f})")
+            elif key == ord('1'):
+                enhance_params['sharp_strength'] = min(3.0, enhance_params['sharp_strength'] + 0.1)
+                print(f"锐化强度: {enhance_params['sharp_strength']:.1f}")
+            elif key == ord('2'):
+                enhance_params['sharp_strength'] = max(0.5, enhance_params['sharp_strength'] - 0.1)
+                print(f"锐化强度: {enhance_params['sharp_strength']:.1f}")
+            elif key == ord('3'):
+                enhance_params['contrast_alpha'] = min(3.0, enhance_params['contrast_alpha'] + 0.1)
+                print(f"对比度: {enhance_params['contrast_alpha']:.1f}")
+            elif key == ord('4'):
+                enhance_params['contrast_alpha'] = max(0.5, enhance_params['contrast_alpha'] - 0.1)
+                print(f"对比度: {enhance_params['contrast_alpha']:.1f}")
             
-    print(detections)
+            frame_count += 1
+            if frame_count % 30 == 0:  # 每30帧打印一次
+                print(f"📊 已处理 {frame_count} 帧")
+                # 打印有检测到的目标
+                detected_items = {k: v for k, v in detections.items() if v is not None and v != []}
+                if detected_items:
+                    print(f"   检测到: {detected_items}")
     
-cv2.destroyAllWindows()
+    except KeyboardInterrupt:
+        print("\n\n用户中断测试")
+    except Exception as e:
+        print(f"\n❌ 测试过程中发生错误: {e}")
+        import traceback
+        traceback.print_exc()
+    finally:
+        # 清理资源
+        get_image.release()
+        cv2.destroyAllWindows()
+        print("🔚 测试结束")
+
+
+if __name__ == "__main__":
+    main()

yolotest2.py

@@ -0,0 +1,227 @@
+"""
+从main.py提取的YOLO识别测试文件
+使用与main.py相同的识别逻辑
+"""
+import cv2
+from utils.get_image import GetImage
+from ultralytics import YOLO
+from config import config_manager
+import os
+
+# 检查模型文件是否存在
+model_path = r"best.pt"
+model0_path = r"best0.pt"
+
+if not os.path.exists(model_path):
+    print(f"❌ 模型文件不存在: {model_path}")
+    exit(1)
+if not os.path.exists(model0_path):
+    print(f"❌ 模型文件不存在: {model0_path}")
+    exit(1)
+
+# 加载YOLO模型（与main.py保持一致）
+try:
+    model = YOLO(model_path).to('cuda')
+    model0 = YOLO(model0_path).to('cuda')
+    print(f"✅ 模型加载成功: {model_path}")
+    print(f"✅ 模型加载成功: {model0_path}")
+except Exception as e:
+    print(f"❌ 模型加载失败: {e}")
+    exit(1)
+
+
+def yolo_shibie(im_PIL, detections, model):
+    """
+    YOLO识别函数（与main.py中的实现完全一致）
+    :param im_PIL: PIL图像对象
+    :param detections: 检测结果字典
+    :param model: YOLO模型
+    :return: 更新后的detections字典
+    """
+    results = model(im_PIL)  # 目标检测
+    for result in results:
+        for i in range(len(result.boxes.xyxy)):
+            left, top, right, bottom = result.boxes.xyxy[i]
+            scalar_tensor = result.boxes.cls[i]
+            value = scalar_tensor.item()
+            label = result.names[int(value)]
+            if label == 'center' or label == 'next' or label == 'boss' or label == 'zhaozi':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(top + (bottom - top) / 2) + 30
+                RW = [player_x, player_y]
+                detections[label] = RW
+            elif label == 'daojv' or label == 'gw':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(top + (bottom - top) / 2) + 30
+                RW = [player_x, player_y]
+                detections[label].append(RW)
+            elif label == 'npc1' or label == 'npc2' or label == 'npc3' or label == 'npc4':
+                player_x = int(left + (right - left) / 2)
+                player_y = int(bottom) + 30
+                RW = [player_x, player_y]
+                detections[label] = RW
+    return detections
+
+
+def main():
+    """主函数"""
+    print("="*60)
+    print("YOLO识别测试（main.py逻辑）")
+    print("="*60)
+    
+    # 从配置加载采集卡设置
+    active_group = config_manager.get_active_group()
+    
+    if active_group is None:
+        print("⚠️ 没有活动的配置组，使用默认设置")
+        print("提示: 可以运行 python gui_config.py 设置配置")
+        cam_index = 0
+        width = 1920
+        height = 1080
+        use_model = model  # 默认使用model
+    else:
+        print(f"📋 使用配置组: {active_group['name']}")
+        cam_index = active_group['camera_index']
+        width = active_group['camera_width']
+        height = active_group['camera_height']
+        use_model = model0  # 城镇中使用model0
+        print(f"   使用模型: model0 (best0.pt) - 用于城镇识别")
+    
+    print(f"   采集卡索引: {cam_index}")
+    print(f"   分辨率: {width}x{height}")
+    print()
+    
+    # 初始化采集卡
+    print("🔧 正在初始化采集卡...")
+    get_image = GetImage(
+        cam_index=cam_index,
+        width=width,
+        height=height
+    )
+    
+    if get_image.cap is None:
+        print("❌ 采集卡初始化失败")
+        print("请检查：")
+        print("1. 采集卡是否正确连接")
+        print("2. 采集卡索引是否正确")
+        print("3. 采集卡驱动是否安装")
+        return
+    
+    print("✅ 采集卡初始化成功")
+    print("\n快捷键:")
+    print("  'q' 或 ESC - 退出")
+    print("  'm' - 切换模型 (model/model0)")
+    print("  'd' - 显示/隐藏检测信息")
+    print()
+    
+    try:
+        frame_count = 0
+        show_detections = True  # 是否显示检测信息
+        current_model = use_model  # 当前使用的模型
+        current_model_name = "model0" if use_model == model0 else "model"
+        
+        while True:
+            # 获取帧
+            frame_data = get_image.get_frame()
+            
+            if frame_data is None:
+                print("⚠️ 无法获取帧，跳过...")
+                continue
+            
+            # frame_data 是 [im_opencv_rgb, im_PIL] 格式
+            im_opencv_rgb, im_PIL = frame_data
+            
+            if im_PIL is None:
+                print("⚠️ PIL图像为空，跳过...")
+                continue
+            
+            # 初始化检测结果字典
+            detections = {
+                'center': None, 'next': None,
+                'npc1': None, 'npc2': None, 'npc3': None, 'npc4': None,
+                'boss': None, 'zhaozi': None,
+                'daojv': [], 'gw': []
+            }
+            
+            # 执行YOLO检测（使用main.py的逻辑）
+            detections = yolo_shibie(im_PIL, detections, current_model)
+            
+            # 获取绘制好框的图像用于显示
+            try:
+                results = current_model(im_PIL)
+                result = results[0]
+                frame_with_boxes_rgb = result.plot()
+                frame_with_boxes_bgr = cv2.cvtColor(frame_with_boxes_rgb, cv2.COLOR_RGB2BGR)
+            except Exception as e:
+                print(f"⚠️ 绘制检测框失败: {e}")
+                frame_with_boxes_bgr = cv2.cvtColor(im_opencv_rgb, cv2.COLOR_RGB2BGR)
+            
+            # 在图像上显示检测信息
+            if show_detections:
+                # 显示模型名称
+                cv2.putText(frame_with_boxes_bgr, f"Model: {current_model_name}", 
+                           (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 2)
+                
+                # 显示检测到的目标
+                y_offset = 60
+                detected_items = []
+                for key, value in detections.items():
+                    if value is not None and value != []:
+                        if key in ['daojv', 'gw']:
+                            detected_items.append(f"{key}: {len(value)}个")
+                        else:
+                            detected_items.append(f"{key}: {value}")
+                
+                if detected_items:
+                    text = f"Detected: {', '.join(detected_items[:5])}"  # 最多显示5个
+                    if len(detected_items) > 5:
+                        text += f" ... (+{len(detected_items)-5})"
+                    cv2.putText(frame_with_boxes_bgr, text, 
+                               (10, y_offset), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (255, 255, 0), 2)
+            
+            # 显示图像
+            cv2.imshow("YOLO Detection (main.py logic)", frame_with_boxes_bgr)
+            
+            # 检查按键
+            key = cv2.waitKey(1) & 0xFF
+            if key in [27, ord('q'), ord('Q')]:
+                print("\n用户退出")
+                break
+            elif key == ord('m') or key == ord('M'):
+                # 切换模型
+                if current_model == model:
+                    current_model = model0
+                    current_model_name = "model0"
+                else:
+                    current_model = model
+                    current_model_name = "model"
+                print(f"切换模型: {current_model_name}")
+            elif key == ord('d') or key == ord('D'):
+                show_detections = not show_detections
+                print(f"显示检测信息: {'开启' if show_detections else '关闭'}")
+            
+            frame_count += 1
+            if frame_count % 30 == 0:  # 每30帧打印一次
+                print(f"📊 已处理 {frame_count} 帧 (模型: {current_model_name})")
+                # 打印有检测到的目标
+                detected_items = {k: v for k, v in detections.items() if v is not None and v != []}
+                if detected_items:
+                    print(f"   检测到: {detected_items}")
+    
+    except KeyboardInterrupt:
+        print("\n\n用户中断测试")
+    except Exception as e:
+        print(f"\n❌ 测试过程中发生错误: {e}")
+        import traceback
+        traceback.print_exc()
+    finally:
+        # 清理资源
+        get_image.release()
+        cv2.destroyAllWindows()
+        print("🔚 测试结束")
+
+
+if __name__ == "__main__":
+    main()
+
+