【opencv450-samples】calcOpticalFlowFarneback计算光流-移动像素的移动方向

#include "opencv2/video/tracking.hpp"
#include "opencv2/imgproc.hpp"
#include "opencv2/videoio.hpp"
#include "opencv2/highgui.hpp"

#include <iostream>

using namespace cv;
using namespace std;

static void help(char** argv)
{
    cout <<
        "\nThis program demonstrates dense optical flow algorithm by Gunnar Farneback\n"
        "Mainly the function: calcOpticalFlowFarneback()\n"
        "Call:\n"
        << argv[0]
        << "This reads from video camera 0\n" << endl;
}
//没step个像素 绘制一次。  cflowmap:上一彩色图像    flow：当前图像相对上一图像的光流
static void drawOptFlowMap(const Mat& flow, Mat& cflowmap, int step,
    double, const Scalar& color)
{
    for (int y = 0; y < cflowmap.rows; y += step)
        for (int x = 0; x < cflowmap.cols; x += step)
        {
            const Point2f& fxy = flow.at<Point2f>(y, x);
            line(cflowmap, Point(x, y), Point(cvRound(x + fxy.x), cvRound(y + fxy.y)),
                color);//光流直线
            circle(cflowmap, Point(x, y), 2, color, -1);//上一图像上的离散点
        }
}

int main(int argc, char** argv)
{
    cv::CommandLineParser parser(argc, argv, "{help h||}{video|balltest.mp4|}");
    if (parser.has("help"))
    {
        help(argv);
        return 0;
    }
    //VideoCapture cap(0);
    VideoCapture cap(samples::findFile(parser.get<String>("video")));
    help(argv);
    if (!cap.isOpened())
        return -1;

    Mat flow, cflow, frame;
    UMat gray, prevgray, uflow;
    namedWindow("flow", 1);

    for (;;)
    {
        cap >> frame;
        if (frame.empty())
        {
            break;
        }
        cvtColor(frame, gray, COLOR_BGR2GRAY);

        if (!prevgray.empty())//上一灰度图非空
        {   //利用用Gunnar Farneback的算法计算全局性的稠密光流算法（即图像上所有像素点的光流都计算出来）
            /// <summary>
            /// /
            /// </summary>
            /// prevgray : 输入前一帧图像
            /// gray : 输入后一帧图像
            /// uflow: flow 输出的光流
            /// 0.5:pyr_scale 金字塔上下两层之间的尺度关系
            /// 3:levels ：金字塔层数
            /// 15: winsize 均值窗口大小，越大越能denoise并且能够检测快速移动目标，但会引起模糊运动区域
            /// 3:iterations 迭代次数
            /// 5: poly_n 像素领域大小，一般为5，7等
            /// 1.2 : poly_sigma 高斯标注差，一般为1-1.5
            /// 0: flags 计算方法。主要包括OPTFLOW_USE_INITIAL_FLOW和OPTFLOW_FARNEBACK_GAUSSIAN
            /// <returns></returns>
            /*CV_EXPORTS_W void calcOpticalFlowFarneback(InputArray prev, InputArray next, InputOutputArray flow,
                double pyr_scale, int levels, int winsize,
                int iterations, int poly_n, double poly_sigma,
                int flags);*/
            calcOpticalFlowFarneback(prevgray, gray, uflow, 0.5, 3, 15, 3, 5, 1.2, 0);//计算光流
            cvtColor(prevgray, cflow, COLOR_GRAY2BGR);//上一灰度图的彩色图
            uflow.copyTo(flow);//光流
            drawOptFlowMap(flow, cflow, 12, 1.5, Scalar(0, 255, 0));//在上一彩色图上绘制绿色光流图
            imshow("flow", cflow);
        }
        if (waitKey(30) >= 0)
            break;
        std::swap(prevgray, gray);//更新上一灰度图
    }
    return 0;
}

运行效果

参考：

OpenCV中CalcOpticalFlowFarneback()函数分析_Dream_yz的博客-CSDN博客https://blog.csdn.net/yzhang6_10/article/details/51225545

OpenCV 光流 | D栈 - Delft Stackhttps://www.delftstack.com/zh/howto/python/opencv-optical-flow/