基于OpenCV实现的视频车辆测速C++源码

/* 基于高斯混合建模(MOG2)+光流法，实现车辆测速 */ #include <iostream> #include <sstream> #include <string.h> #include <stdio.h> #include <opencv2/opencv.hpp> #include <opencv2/core/core.hpp> #include <opencv2/imgproc/imgproc.hpp> #include <opencv2/video/video.hpp> #include <opencv2/highgui/highgui.hpp> #include <opencv2/objdetect/objdetect.hpp> #include <opencv2/video/tracking.hpp> #include <cmath> #include <vector> #include <typeinfo> #include <opencv2/video/background_segm.hpp> #include <time.h> using namespace std; using namespace cv; /*计算车速*/ double get_speed(cv::Mat& frame, int index, vector<Point>& contour, Point2f center, float radius, vector<Point2f> prevContourCenters, double fps) { int meters = 30; int maxSpeed = 100; if (center.x > 5 && center.x < 480){ for (size_t i = 0; i < prevContourCenters.size(); i++) { //if (prevContourCenters[i].x < center.x){ if (prevContourCenters[i].y < center.y) { //double diff = center.x - prevContourCenters[i].x; double diff = center.y - prevContourCenters[i].y; cv::Point x, y; int fontFace = 3; double fontScale = 0.5; int thickness = 1.8; Point textOrg(center.x - 20, center.y - radius - 5); String str2 = " km/h"; double speed = diff*fps / 300 * meters*3.6; if (speed > 100) speed = 101; std::stringstream strSpeed; strSpeed << speed; if (speed > maxSpeed){ //在图像上显示车速,保存超速车辆图片 cv::putText(frame, strSpeed.str()+str2, textOrg, fontFace, fontScale, Scalar(0, 255, 0), thickness, 5); imwrite("image.jpg", frame); } else { cout << "Speed =" << strSpeed.str() + str2 << endl; cv::putText(frame, strSpeed.str() + str2, textOrg, fontFace, fontScale, Scalar(0, 255, 0), thickness, 5); } return speed; } } } } int main(int argc, char *argv[]) { //车辆数量 int count; //车辆轮廓面积、宽高比 double area, ar; //当前处理帧,未处理前的图像 Mat frame; //高斯混合模型分割出的前景 Mat fore; //当前处理帧(双边滤波后图像) Mat img; //当前帧的上一帧图像 Mat prevImg; //灰度图像 Mat gray; //灰度图像转换成CV_8UC1类型 Mat temp; //保存车辆ROI图像 Mat vehicle_ROI; //获取到视频的第一帧原始图像 Mat img_temp; //打开视频文件 VideoCapture cap("test.mp4"); if (!cap.isOpened()) { cout << "打开视频文件失败" << endl; return -1; } //创建高斯混合模型，用于检测视频中的移动目标(车辆) Ptr<BackgroundSubtractorMOG2> bg = createBackgroundSubtractorMOG2(500, 25, false); vector<vector<Point> > contours; vector<Point2f> prevContoursCenters; vector<Point2f> tempContoursCenters; vector<Rect> cars; //获取第一帧 cap >> img_temp; //RGB转灰度化 cvtColor(img_temp, gray, CV_BGR2GRAY); //转换成CV_8UC1类型 gray.convertTo(temp, CV_8U); //双边滤波,可以滤除图像数据中的噪声 bilateralFilter(temp, prevImg, 5, 20, 20); //开始和结束时间，用于计时 time_t start, end; // 帧率，表示帧数/秒 double fps; // 当前帧数 int counter = 0; // 从开始到当前时间的秒数，用浮点数表示 double sec; //开始计时 time(&start); while (true) { count = 0; cap >> frame; if (frame.empty()) break; //imshow("Frame", frame); //waitKey(2); //灰度化 cvtColor(frame, gray, CV_BGR2GRAY); //转换为CV_8UC1类型 gray.convertTo(temp, CV_8U); //双边滤波 bilateralFilter(temp, img, 5, 20, 20); //计算前景mask bg->apply(img, fore); //腐蚀图片 erode(fore, fore, Mat()); //膨胀、扩大图片 dilate(fore, fore, Mat()); //查找轮廓 findContours(fore, contours, CV_RETR_EXTERNAL, CV_CHAIN_APPROX_SIMPLE); vector<vector<Point> > contours_poly(contours.size()); vector<Rect> boundRect(contours.size()); vector<Point2f> center(contours.size()); vector<float> radius(contours.size()); tempContoursCenters.clear(); time(&end); //计算当前帧率 ++counter; //开始到结束经历的时间 sec = difftime(end, start); //帧率=帧数/时间 fps = counter / sec; for (size_t i = 0; i < contours.size(); i++) { //对轮廓进行多边形拟合 approxPolyDP(Mat(contours[i]), contours_poly[i], 10, true); //最小外接矩形 boundRect[i] = boundingRect(Mat(contours_poly[i])); //保存车辆ROI图像 vehicle_ROI = img(boundRect[i]); //计算车辆轮廓面积 area = contourArea(contours[i], false); //宽高比 ar = vehicle_ROI.cols / vehicle_ROI.rows; //面积过滤，将符合阈值范围内的车辆轮廓绘制出来并计算出车速值 if (area > 450.0 && ar > 0.8) { //最小外接圆,求出圆心坐标和半径 minEnclosingCircle((Mat)contours_poly[i], center[i], radius[i]); //绘制矩形 rectangle(frame, boundRect[i].tl(), boundRect[i].br(), Scalar(0, 255, 0), 1, 8, 0); tempContoursCenters.push_back(center[i]); //计算出车速值 get_speed(frame, i, contours[i], center[i], radius[i], prevContoursCenters, fps); count = count + 1; } } stringstream ss; ss << count; //显示内容 string s = ss.str(); //字体类型 int fontFace = FONT_HERSHEY_SCRIPT_SIMPLEX; double fontScale = 1; //字体大小 int thickness = 3; //字体粗细 cv::Point textOrg(10, 110); cv::putText(frame, s, textOrg, fontFace, fontScale, Scalar(255, 255, 255), thickness, 5); int win_size = 10; int maxCorners = 200; double qualityLevel = 0.5; double minDistance = 1; int blockSize = 3; double k = 0.04; vector<Point2f> img_corners; img_corners.reserve(maxCorners); vector<Point2f> prevImg_corners; prevImg_corners.reserve(maxCorners); //角点检测，获得像素精度角点坐标 goodFeaturesToTrack(img, img_corners, maxCorners, qualityLevel, minDistance, Mat(), blockSize, true); //获得亚像素精度角点坐标 cornerSubPix(img, img_corners, Size(win_size, win_size), Size(-1, -1), TermCriteria(CV_TERMCRIT_ITER | CV_TERMCRIT_EPS, 20, 0.03)); vector<uchar> features_found; features_found.reserve(maxCorners); vector<float> feature_errors; feature_errors.reserve(maxCorners); //Lukas-Kanade光流，用于车辆跟踪 calcOpticalFlowPyrLK(img, prevImg, img_corners, prevImg_corners, features_found, feature_errors, Size(win_size, win_size), 3, cvTermCriteria(CV_TERMCRIT_ITER | CV_TERMCRIT_EPS, 20, 0.3), 0, k); // for( int i=0; i < features_found.size(); i++ ){ // Point2f p0( ceil( img_corners[i].x ), ceil( img_corners[i].y ) ); // Point2f p1( ceil( prevImg_corners[i].x ), ceil( prevImg_corners[i].y ) ); // line( frame, p0, p1, CV_RGB(255,0,0), 5 ); // } prevImg = img; prevContoursCenters.clear(); prevContoursCenters = tempContoursCenters; //显示结果 namedWindow("Frame"); imshow("Frame", frame); if (waitKey(1) >= 0) break; } //销毁所有窗口 destroyAllWindows(); system("PAUSE"); }