本篇讲解opencv video鼠标选中的物体跟踪，使用的是opencv提供的calcOpticalFlowPyrLK。
void calcOpticalFlowPyrLK(InputArray prevImg, InputArray nextImg, InputArray prevPts, InputOutputArray nextPts,
OutputArray status, OutputArray err, Size winSize=Size(21,21), int maxLevel=3,
TermCriteria criteria=TermCriteria(TermCriteria::COUNT+TermCriteria::EPS, 30, 0.01),
int flags=0, double minEigThreshold=1e-4 );
prevImg:前一帧video图像。
nextImg:当前video图像。
prevPts:前一帧video图像中被跟踪的坐标点。
nextPts：prevPts保存的坐标点，在当前帧video图像中计算出来的对应坐标，也就是跟踪到的坐标点。
winSize：在每层的搜索窗口的大小。
criteria：算法递归停止的条件。
。。。。。
#include "opencv2/video/tracking.hpp" #include "opencv2/imgproc/imgproc.hpp" #include "opencv2/highgui/highgui.hpp" #include <iostream> #include <ctype.h> #include <stdio.h> #include <unistd.h> #include <stdlib.h> using namespace cv; using namespace std; vector<Point2f> point1, point2; bool left_mouse = false; Point2f point; int pic_info[4]; Mat gray, prevGray, image; const Scalar GREEN = Scalar(0,255,0); int rect_width = 0, rect_height = 0; Point tmpPoint; static void onMouse( int event, int x, int y, int /*flags*/, void* /*param*/ ){ Mat mouse_show; image.copyTo(mouse_show); if(event == CV_EVENT_LBUTTONDOWN){ pic_info[0] = x; pic_info[1] = y; left_mouse = true; }else if(event == CV_EVENT_LBUTTONUP){ rectangle(mouse_show, Point(pic_info[0], pic_info[1]), Point(x, y), GREEN, 2); rect_width = <a href="http://www.opengroup.org/onlinepubs/%3Cspan%20class=" nu19"="" style="text-decoration: none; color: rgb(11, 0, 128); background-image: none; background-position: initial initial; background-repeat: initial initial;">009695399/functions/abs.html">abs(x - pic_info[0]); rect_height = <a href="http://www.opengroup.org/onlinepubs/%3Cspan%20class=" nu19"="" style="text-decoration: none; color: rgb(11, 0, 128); background-image: none; background-position: initial initial; background-repeat: initial initial;">009695399/functions/abs.html">abs(y - pic_info[1]); x = (pic_info[0] + x) / 2; y = (pic_info[1] + y) / 2; point = Point2f((float)x, (float)y); point1.clear(); point2.clear(); point1.push_back(point); imshow("LK Demo", mouse_show); left_mouse = false; }else if((event == CV_EVENT_MOUSEMOVE) && (left_mouse == true)){ rectangle(mouse_show, Point(pic_info[0], pic_info[1]), Point(x, y), GREEN, 2); imshow("LK Demo", mouse_show); } } int main( int argc, char** argv ) { VideoCapture cap; TermCriteria termcrit(CV_TERMCRIT_ITER|CV_TERMCRIT_EPS, 20, 0.03); //迭代算法的终止条件 Size winSize(31,31); cap.open(argv[1]); if(!cap.isOpened()){ cout << "Could not initialize capturing...\n"; return 0; } namedWindow( "LK Demo", 1 ); setMouseCallback( "LK Demo", onMouse, 0 ); for(;;){ Mat frame; cap >> frame; if( frame.empty() ) break; frame.copyTo(image); cvtColor(image, gray, COLOR_BGR2GRAY); if((!point1.empty())){ vector<uchar> status; vector<float> err; if(prevGray.empty()) gray.copyTo(prevGray); calcOpticalFlowPyrLK(prevGray, gray, point1, point2, status, err, winSize, 3, termcrit, 0, 0.001); //使用金字塔Lucas&Kanade方法计算一个稀疏特征集的光流 tmpPoint = point2[0]; rectangle(image, Point(tmpPoint.x - 20, tmpPoint.y - 20), Point(tmpPoint.x + 20, tmpPoint.y + 20), GREEN, 2); } imshow("LK Demo", image); waitKey(100); std::swap(point2, point1); cv::swap(prevGray, gray); } return 0; }
1、首先设置了算法calcOpticalFlowPyrLK将会使用到的递归停止条件(termcrit),关于termcrit的具体讲解，可以看这里有具体讲解: http://blog.csdn.net/yang_xian521/article/details/6905244 ，接着打开视频文件，句柄保存在cap中。然后设置了显示窗口，已经它的鼠标响应函数。
VideoCapture cap; TermCriteria termcrit(CV_TERMCRIT_ITER|CV_TERMCRIT_EPS, 20, 0.03); //迭代算法的终止条件 Size winSize(31,31); cap.open(argv[1]); if(!cap.isOpened()){ cout << "Could not initialize capturing...\n"; return 0; } namedWindow( "LK Demo", 1 ); setMouseCallback( "LK Demo", onMouse, 0 );
2、鼠标响应函数，主要做的就是，在当前video帧中画一个矩形，然后计算出该矩形的中心位置坐标,保存到point1中。这个位置坐标就是在 calcOpticalFlowPyrLK算法中用来跟踪的点。
static void onMouse( int event, int x, int y, int /*flags*/, void* /*param*/ ){ Mat mouse_show; image.copyTo(mouse_show); if(event == CV_EVENT_LBUTTONDOWN){ pic_info[0] = x; pic_info[1] = y; left_mouse = true; }else if(event == CV_EVENT_LBUTTONUP){ rectangle(mouse_show, Point(pic_info[0], pic_info[1]), Point(x, y), GREEN, 2); rect_width = <a href="http://www.opengroup.org/onlinepubs/%3Cspan%20class=" nu19"="" style="text-decoration: none; color: rgb(11, 0, 128); background-image: none; background-position: initial initial; background-repeat: initial initial;">009695399/functions/abs.html">abs(x - pic_info[0]); rect_height = <a href="http://www.opengroup.org/onlinepubs/%3Cspan%20class=" nu19"="" style="text-decoration: none; color: rgb(11, 0, 128); background-image: none; background-position: initial initial; background-repeat: initial initial;">009695399/functions/abs.html">abs(y - pic_info[1]); x = (pic_info[0] + x) / 2; y = (pic_info[1] + y) / 2; point = Point2f((float)x, (float)y); point1.clear(); point2.clear(); point1.push_back(point); imshow("LK Demo", mouse_show); left_mouse = false; }else if((event == CV_EVENT_MOUSEMOVE) && (left_mouse == true)){ rectangle(mouse_show, Point(pic_info[0], pic_info[1]), Point(x, y), GREEN, 2); imshow("LK Demo", mouse_show); } }
3、当用户还没有鼠标框选跟踪目标时候，软件会不断的读取出video的数据，保存到frame中，接着copy一份当前帧数据到gray中，并将gray中的 图像灰阶化，然后显示出video frame数据。最后交换了point2和point1中的坐标信息和保存了当前灰阶化后的帧率到prevGray中。
for(;;){ Mat frame; cap >> frame; if( frame.empty() ) break; frame.copyTo(image); cvtColor(image, gray, COLOR_BGR2GRAY); ........... imshow("LK Demo", image); waitKey(100); std::swap(point2, point1); cv::swap(prevGray, gray); }
4、最后当用户框选了跟踪目标之后，也就是point1不为空之后，开始用calcOpticalFlowPyrLK跟踪计算，注意传入该函数的参数:prevGray相当于 之前保存的前一帧的数据；gray是当前帧数据；point1是前一帧中被跟踪的目标位置；point2是计算出来的被跟踪目标在当前帧的位置。 最后用计算出来的在当前帧中，跟踪目标坐标point2作为中心，在当前帧中画出一个40X40的矩形作为标记，最后显示出来。
if((!point1.empty())){ vector<uchar> status; vector<float> err; if(prevGray.empty()) gray.copyTo(prevGray); calcOpticalFlowPyrLK(prevGray, gray, point1, point2, status, err, winSize, 3, termcrit, 0, 0.001); //使用金字塔Lucas&Kanade方法计算一个稀疏特征集的光流 tmpPoint = point2[0]; rectangle(image, Point(tmpPoint.x - 20, tmpPoint.y - 20), Point(tmpPoint.x + 20, tmpPoint.y + 20), GREEN, 2); }
对应的效果演示如下:
原文地址：http://blog.csdn.net/u011630458/article/details/45950319
