正如上一篇文章所述， cvCalcOpticalFlowPyrLK 需要确定特征点。
本程序，通过帧差获得运动的点作为特征点。
本程序原本的目的是计算运动点的速度，通过修正可以进行运动跟踪。
#include <cv.h>
#include <highgui.h>
#include <iostream>
using namespace std;
int const MAX_CORNERS = 1000;
int main (int argc, char **argv)
{
CvCapture* capture = 0;
capture = cvCaptureFromCAM( CV_CAP_ANY ); //get frame
IplImage *src_img1; //the previous frame (gray)
IplImage *src_img2; //the current frame(gray)
IplImage *dst_img; //the result
IplImage *cur_img;
IplImage *pre_img;
CvPoint2D32f * move_old_point = new CvPoint2D32f[ MAX_CORNERS];
CvPoint2D32f * move_new_point = new CvPoint2D32f[ MAX_CORNERS];
char *features_found = new char[MAX_CORNERS];
float *features_error = new float[MAX_CORNERS];
CvTermCriteria criteria;
criteria = cvTermCriteria (CV_TERMCRIT_ITER | CV_TERMCRIT_EPS, 64, 0.01);
while(1)
{
int i,j;
int dx, dy;
int p = 0;
int rows, cols;
int countn = MAX_CORNERS;
pre_img = cvQueryFrame(capture);
CvSize img_sz = cvGetSize(pre_img);
src_img1 = cvCreateImage(img_sz, IPL_DEPTH_8U, 1);
cvCvtColor(pre_img, src_img1, CV_RGB2GRAY);
cur_img = cvQueryFrame(capture);
src_img2 = cvCreateImage(img_sz, IPL_DEPTH_8U, 1);
cvCvtColor(cur_img, src_img2, CV_RGB2GRAY);
dst_img = (IplImage *)cvClone(cur_img);
IplImage *move_img = cvCreateImage(img_sz, IPL_DEPTH_8U, 1);
cvZero(move_img);
//cvAbsDiff(src_img1, src_img2,move_img);
cols = src_img1->width;
rows = src_img1->height;
for (i = 0; i <cols; i++)
{
for (j = 0; j<rows; j++)
{
double a = abs(cvGet2D(src_img1, j, i).val[0]-cvGet2D(src_img2, j, i).val[0]);
CvScalar b = cvScalar(a, 0, 0,0);
cvSet2D(move_img, j, i,b);
if (a>40)
{
if (p<MAX_CORNERS-1)
{
int d = ++p;
move_old_point[d].x = i;
move_old_point[d].y = j;
}
}
}
}
cvNamedWindow("moving object", 1);
cvShowImage("moving object", move_img);
CvSize Pyrsize = cvSize(src_img1->width +8, src_img1->height/3);
IplImage * pyrA = cvCreateImage(Pyrsize, IPL_DEPTH_32F, 1); //pyrA是需要寻找的点，不是没有初始化的
IplImage * pyrB = cvCreateImage(Pyrsize, IPL_DEPTH_32F, 1);
cvCalcOpticalFlowPyrLK(src_img1,
src_img2,
pyrA,
pyrB,
move_old_point,
move_new_point,
countn,
cvSize(10, 10),
3,
features_found,
features_error,
criteria,
0
);
for (i = 0; i < countn; i++)
{
int x1 = (int)move_new_point[i].x;
int x2 = (int)move_old_point[i].x;
int y1 = (int)move_new_point[i].y;
int y2 = (int)move_old_point[i].y;
dx =(int) abs(x1 - x2) ;
dy = (int)abs(y1 - y2);
if (dx >= 5&& dy >= 5)
{
cvLine (dst_img, cvPoint(x2, y2),cvPoint(x2+5, y2+5) , CV_RGB (255, 0, 0), 1, CV_AA, 0);
}
}
cvNamedWindow ("ImagePyrLK", 1);
cvShowImage ("ImagePyrLK", dst_img);
cvWaitKey (1);
cvReleaseImage (&dst_img);
cvReleaseImage(&pyrA);
cvReleaseImage(&pyrB);
cvReleaseImage(&move_img);
}
cvDestroyWindow("moving object");
cvDestroyWindow ("ImagePyrLK");
cvReleaseImage (&src_img1);
cvReleaseImage (&src_img2);
cvReleaseImage (&pre_img);
cvReleaseImage (&cur_img);
return 0;
}
