簡介

　　本篇講解使用opencv提供的流光法演算法介面，實現物體跟蹤。範例程式碼為參考修改tvl1_optical_flow.cpp實現。

具體實現

實現程式碼

#include <iostream>
#include <fstream>
 
#include "opencv2/video/tracking.hpp"
#include "opencv2/highgui/highgui.hpp"
 
using namespace cv;
using namespace std;
 
inline bool isFlowCorrect(Point2f u)
{
    return
 
 !cvIsNaN(u.x) && !cvIsNaN(u.y) && fabs(u.y) < 1e9;
}
 
static Vec3b computeColor(float fx, float fy)
{
    static bool first = true;
 
    // relative lengths of color transitions:
    // these are chosen based on perceptual similarity
    // (e.g. one can distinguish more shades between red and yellow
 

    //  than between yellow and green)
    const int RY = 15;
    const int YG = 6;
    const int GC = 4;
    const int CB = 11;
    const int BM = 13;
    const int MR = 6;
    const int NCOLS = RY + YG + GC + CB + BM + MR;
    static Vec3i colorWheel[NCOLS];
 
    if (first){
        int k = 0;
 
        for
 
 (int i = 0; i < RY; ++i, ++k)
            colorWheel[k] = Vec3i(255, 255 * i / RY, 0);
 
        for (int i = 0; i < YG; ++i, ++k)
            colorWheel[k] = Vec3i(255 - 255 * i / YG, 255, 0);
 
        for (int i = 0; i < GC; ++i, ++k)
            colorWheel[k] = Vec3i(0, 255, 255 * i / GC);
 
        for (int i = 0; i < CB; ++i, ++k)
            colorWheel[k] = Vec3i(0, 255 - 255 * i / CB, 255);
 
        for (int i = 0; i < BM; ++i, ++k)
            colorWheel[k] = Vec3i(255 * i / BM, 0, 255);
 
        for (int i = 0; i < MR; ++i, ++k)
            colorWheel[k] = Vec3i(255, 0, 255 - 255 * i / MR);
 
        first = false;
    }
 
    const float rad = sqrt(fx * fx + fy * fy);
    const float a = atan2(-fy, -fx) / (float)CV_PI;
 
    const float fk = (a + 1.0f) / 2.0f * (NCOLS - 1);
    const int k0 = static_cast<int>(fk);
    const int k1 = (k0 + 1) % NCOLS;
    const float f = fk - k0;
 
    Vec3b pix;
 
    for (int b = 0; b < 3; b++)
    {
        const float col0 = colorWheel[k0][b] / 255.f;
        const float col1 = colorWheel[k1][b] / 255.f;
 
        float col = (1 - f) * col0 + f * col1;
 
        if (rad <= 1)
            col = 1 - rad * (1 - col); // increase saturation with radius
        else
            col *= .75; // out of range
 
        pix[2 - b] = static_cast<uchar>(255.f * col);
    }
 
    return pix;
}
 
static void drawOpticalFlow(const Mat_<Point2f>& flow, Mat& dst, float maxmotion = -1)
{
    dst.create(flow.size(), CV_8UC3);
    dst.setTo(Scalar::all(0));
 
    // determine motion range:
    float maxrad = maxmotion;
 
    if (maxmotion <= 0)
    {
        maxrad = 1;
        for (int y = 0; y < flow.rows; ++y)
        {
            for (int x = 0; x < flow.cols; ++x)
            {
                Point2f u = flow(y, x);
 
                if (!isFlowCorrect(u))
                    continue;
 
                maxrad = max(maxrad, sqrt(u.x * u.x + u.y * u.y));
            }
        }
    }
 
    for (int y = 0; y < flow.rows; ++y)
    {
        for (int x = 0; x < flow.cols; ++x)
        {
            Point2f u = flow(y, x);
 
            if (isFlowCorrect(u))
                dst.at<Vec3b>(y, x) = computeColor(u.x / maxrad, u.y / maxrad);
        }
    }
}
 
int main(int argc, const char* argv[])
{
	Mat frame0;
	Mat frame1;
	Mat_<Point2f> flow;
	Ptr<DenseOpticalFlow> tvl1 = createOptFlow_DualTVL1();
	Mat out;
 
	if (argc < 2){
		cerr << "Usage : " << argv[0] << "<video>" << endl;
		return -1;
	}
	VideoCapture cap;
	cap.open(argv[1]);
 
	while(1){
		cap >> frame0;
		if(frame0.empty()){
			cerr<< "video is over!!" << endl;
			break;
		}
		cvtColor(frame0, frame0, CV_BGR2GRAY);
		if(!frame1.empty()){
			const double start = (double)getTickCount();
			tvl1->calc(frame0, frame1, flow);
			const double timeSec = (getTickCount() - start) / getTickFrequency();
			cout << "calcOpticalFlowDual_TVL1 : " << timeSec << " sec" << endl;
			drawOpticalFlow(flow, out);
			imshow("out", out);
			imshow("src", frame0);
			waitKey(10);
		}
		frame0.copyTo(frame1);
	}
    waitKey();
 
    return 0;
}

程式碼講解

　　1、建立了一個DenseOpticalFlow例項，同時獲得打開了需要跟蹤處理的video視訊到cap中。

        Ptr<DenseOpticalFlow> tvl1 = createOptFlow_DualTVL1();
	Mat out;
 
	if (argc < 2){
		cerr << "Usage : " << argv[0] << "<video>" << endl;
		return -1;
	}
	VideoCapture cap;
	cap.open(argv[1]);

　　2、在迴圈中，不斷的讀取video的幀資料到frame0中，接著cvtColor將frame0中的資料，灰階化。判斷到儲存前一幀資料為空，也就是表示
剛剛讀取到第一幀資料時候，不進入處理函式中，直接跳過。最後將frame0中的幀資料，儲存到frame1中。frame0進入下一次迴圈，獲得新一幀
資料。

     while(1){
	cap >> frame0;
	if(frame0.empty()){
		cerr<< "video is over!!" << endl;
		break;
	}
	cvtColor(frame0, frame0, CV_BGR2GRAY);
	if(!frame1.empty()){
                ...........
                ...........
	}
	frame0.copyTo(frame1);
     }

　　3、當檢測到frame1儲存了前一幀資料之後，進入到流光法計算中。首先獲得當前時鐘getTickCount。使用tvl1->calc分別傳入當前
幀(frame0)和前一幀(frame1)，將獲得的位置偏移儲存到flow中。接著計算出calc函式處理花費的時間，之後使用函式 
drawOpticalFlow，利用flow中的位置偏移，根據偏移位置的方向和速度，從而在out影象，對應位置賦予不同的顏色和飽和度。最後將
當前幀影象和處理之後的out影象分別顯示出來。

        const double start = (double)getTickCount();
	tvl1->calc(frame0, frame1, flow);
	const double timeSec = (getTickCount() - start) / getTickFrequency();
	cout << "calcOpticalFlowDual_TVL1 : " << timeSec << " sec" << endl;
	drawOpticalFlow(flow, out);
	imshow("out", out);
	imshow("src", frame0);
	waitKey(10);

效果演示

　對應的效果演示如下: