机器视觉实用工具集NO.16——指定监控区域运动物体闯入检测工具

前言

在安防领域，很常见的一个机器视觉检测任务是区域闯入报警侦测，实现此功能，并不是非常困难，可以利用光流法对视频固定区域内的图像（相邻帧的）特征点变化情况进行判断，本文利用了opencv中的calcOpticalFlowFarneback稠密光流函数为核心，实现了在监控视频可调四边形区域内，实现了运动物体侦测，并报警。效果：
如下图，当四边形内有运动物体闯入或变化时，选区变成红色报警：
区域正常为绿色：

区域报警：

原理

本文实现的区域运动物体检测实现流程是：
1、光流检测获取变化的像素
2、判断变化的像素是否在手动选择的四边形边框内
3、计算四边形边框内变化的特征像素个数，超过一个限值如50则发出报警

Python实现源码

# -*- coding: utf-8 -*-
"""
Created on Sun Apr 10 09:17:51 2022

@author: JAMES FEI 
Copyright (C) 2022 FEI PANFENG, All rights reserved.
THIS SOFTEWARE, INCLUDING DOCUMENTATION,IS PROTECTED BY COPYRIGHT CONTROLLED 
BY FEI PANFENG ALL RIGHTS ARE RESERVED.
"""


import numpy as np
import cv2

step=10
global points
def onMouse(event, x, y, flags, param):
    global points #声明引用全局变量    
    if event == cv2.EVENT_LBUTTONDOWN: #鼠标点击弹起事件 
        print(x,y)
        r=100000000
        pi=5
        for i in range(len(points)):            
            if r>((x-points[i][0])**2+(y-points[i][1])**2):
                r=(x-points[i][0])**2+(y-points[i][1])**2
                pi=i
        print("point:",pi)
        if pi!=5:            
            points[pi]=[x,y] 
            print(points)
    
def drawRec(img,points,flag):
    """
    画出境界区域
    """
    if flag:
        color=(255,0,255)
    else:
        color=(0,255,0)
        
    for i in range(len(points)):        
        cv2.circle(img, (points[i][0],points[i][1]), 10, color, 2) 
        if i!=3:
            cv2.line(img, (points[i][0],points[i][1]),(points[i+1][0],points[i+1][1]), color, 4)
        else:
            cv2.line(img, (points[i][0],points[i][1]),(points[0][0],points[0][1]),color, 4)
    return img    
def isinRec(point,points):
    """
    是否在边框内
    """

    [x1,y1] = points[0]
    [x2,y2] = points[1]
    [x3,y3] = points[2]
    [x4,y4] = points[3]
    p1 = (x1-x4)*(point[1]-y4)-(y1-y4)*(point[0]-x4)
    p2 = (x2-x1)*(point[1]-y1)-(y2-y1)*(point[0]-x1)
    p3 = (x3-x2)*(point[1]-y2)-(y3-y2)*(point[0]-x2)
    p4 = (x4-x3)*(point[1]-y3)-(y4-y3)*(point[0]-x3)
    #print(a,b,c,d)
    if (p1>0 and p2>0 and p3>0 and p4>0) or (p1<0 and p2<0 and p3<0 and p4<0):
        return True
    else:
        return False


if __name__ == '__main__':    
    cam = cv2.VideoCapture(0)
    ret, prev = cam.read()
    H=prev.shape[0]
    W=prev.shape[1]
    points=[
            [50,H-50],
            [50,50],
            [W-50,50],        
            [W-50,H-50]                               
            ]    
    prevgray = cv2.cvtColor(prev, cv2.COLOR_BGR2GRAY)
    cv2.namedWindow('warning') #创建窗口    
    cv2.setMouseCallback('warning',onMouse) #设置窗口鼠标回调函数

    while True:
        ret, frame = cam.read()
        if ret:            
            H=frame.shape[0]
            W=frame.shape[1]
        else:
            H=100
            W=100

        gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
        
        flow = cv2.calcOpticalFlowFarneback(prevgray, gray, None, 0.5, 3, 15, 3, 5, 1.1, 0)
        """
        prevImg – First 8-bit single-channel input image.   输入单通道图片
        nextImg – Second input image of the same size and the same type asprevImg . 下一帧图片。
        flow – Computed flow image that has the same size as prevImg and typeCV_32FC2 .输出的双通道flow
        pyrScale – Parameter specifying the image scale (<1) to build pyramids for each image. pyrScale=0.5 means a classical pyramid, where each next layer is twice smaller than the previous one.金字塔上上下两层之间的尺度关系。
        levels – Number of pyramid layers including the initial image.levels=1 means that no extra layers are created and only the original images are used. 金字塔层数
        winsize – Averaging window size. Larger values increase the algorithm robustness to image noise and give more chances for fast motion detection, but yield more blurred motion field.均值窗口大小，越大越能denoise并且能够检测快速移动目标，但是会引起模糊运动区域。
        iterations – Number of iterations the algorithm does at each pyramid level.迭代次数。
        polyN – Size of the pixel neighborhood used to find polynomial expansion in each pixel. Larger values mean that the image will be approximated with smoother surfaces, yielding more robust algorithm and more blurred motion field. Typically, polyN =5 or 7.
        polySigma – Standard deviation of the Gaussian that is used to smooth derivatives used as a basis for the polynomial expansion. ForpolyN=5 , you can set polySigma=1.1 . For polyN=7 , a good value would be polySigma=1.5
        flag-OPTFLOW_USE_INITIAL_FLOW  OPTFLOW_FARNEBACK_GAUSSIAN
        """
        prevgray = gray
        # 绘制线
        h, w = gray.shape[:2]
        y, x = np.mgrid[step / 2:h:step, step / 2:w:step].reshape(2, -1).astype(int)#等差数列
        fx, fy = flow[y, x].T
        lines = np.vstack([x, y, x + fx, y + fy]).T.reshape(-1, 2, 2)
        lines = np.int32(lines)

        line = []

        movepoint=0        
        
        for l in lines:
            if l[0][0]-l[1][0]>3 or l[0][1]-l[1][1]>3:                
                line.append(l)
                if isinRec(l[0].tolist(),points):
                    movepoint+=1

        if movepoint>50:  
            print(movepoint)
            cv2.polylines(frame, line, 0, (255,255,0))
            pimg=drawRec(frame,points,True)  
        else:
            cv2.polylines(frame, line, 0, (0,255,0))
            pimg=drawRec(frame,points,False)              
              
        cv2.imshow('warning', pimg)
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break

    cv2.destroyAllWindows()