人员闯入危险区域AI智能分析预警系统：从被动监控到主动防御的智能实践

import numpy as np
from typing import Dict, List, Tuple, Optional

class IntrusionDetector3D:
    def __init__(self, camera_matrix: np.ndarray, 
                 ground_plane: Tuple[float, float, float, float],
                 forbidden_zones: List[List[Tuple[float, float]]]):
        """
        :param camera_matrix: 3x3 相机内参矩阵
        :param ground_plane: (a,b,c,d) 地面平面方程 ax+by+cz+d=0
        :param forbidden_zones: 每个禁区由世界坐标点列表定义 [(x1,y1), (x2,y2), ...]
        """
        self.K = camera_matrix
        self.plane = ground_plane
        self.zones = forbidden_zones  # 已转换为世界坐标系的2D多边形

    def image_to_ground(self, pixel_x: int, pixel_y: int) -> Optional[Tuple[float, float]]:
        """将图像像素坐标反投影至地面世界坐标"""
        # 简化假设：目标脚部位于地面上
        u, v = pixel_x, pixel_y
        fx, fy = self.K[0,0], self.K[1,1]
        cx, cy = self.K[0,2], self.K[1,2]
        
        # 射线与地面求交（省略详细推导）
        # 实际应用中需结合人体检测框底部中心点
        # 此处返回模拟值
        world_x = (u - cx) * 2.0 / fx  # 示例映射
        world_y = (v - cy) * 2.0 / fy
        return (world_x, world_y)

    def is_in_forbidden_zone(self, world_point: Tuple[float, float]) -> bool:
        """判断世界坐标点是否在任一禁区内（射线法）"""
        x, y = world_point
        for zone in self.zones:
            n = len(zone)
            inside = False
            p1x, p1y = zone[0]
            for i in range(n + 1):
                p2x, p2y = zone[i % n]
                if y > min(p1y, p2y) and y <= max(p1y, p2y) and x <= max(p1x, p2x):
                    if p1y != p2y:
                        xinters = (y - p1y) * (p2x - p1x) / (p2y - p1y) + p1x
                    else:
                        xinters = p1x
                    if p1x == p2x or x <= xinters:
                        inside = not inside
                p1x, p1y = p2x, p2y
            if inside:
                return True
        return False

    def evaluate_intrusion(self, detection_bbox: List[int], 
                           action_type: str) -> Dict[str, any]:
        """
        :param detection_bbox: [x, y, w, h] 人体检测框（像素）
        :param action_type: "walk"/"climb"/"crawl" 等行为标签
        :return: 入侵评估结果
        """
        # 取检测框底部中心作为脚部位置
        foot_x = detection_bbox[0] + detection_bbox[2] // 2
        foot_y = detection_bbox[1] + detection_bbox[3]
        
        world_pos = self.image_to_ground(foot_x, foot_y)
        if world_pos is None:
            return {"intruded": False, "reason": "projection_failed"}
            
        in_zone = self.is_in_forbidden_zone(world_pos)
        # 结合行为类型提升判定置信度
        high_risk_action = action_type in ["climb", "crawl", "cut_fence"]
        
        return {
            "intruded": in_zone,
            "high_risk": high_risk_action,
            "world_position": world_pos,
            "action": action_type
        }

# 使用示例
# detector = IntrusionDetector3D(camera_matrix=K, ground_plane=(0,0,1,0), 
#                                forbidden_zones=)
# result = detector.evaluate_intrusion([320, 400, 60, 180], "climb")
# if result["intruded"] and result["high_risk"]: trigger_alarm(result)