c++ OpenCV memory management - c++

Question

I'm trying to build a program that recognises objects through template matching.
Whenever I select my ROI the program memory starts to gain 2-3 mbs per second and after 20-30 minutes the program crashes because it uses to much memory.
I think the problem starts somewhere in the code below but I have no idea where, and how to somehow free the memory without corrupting the program.
Could someone give me a lead on how to fix this problem?
void testApp::update()
{
vidGrabber.grabFrame();
if (vidGrabber.isFrameNew())
{
colorImg.setFromPixels(vidGrabber.getPixels(), camWidth,camHeight);
if(subjectIsDefined)
{
IplImage *result = cvCreateImage(cvSize(camWidth - subjectImg.width + 1, camHeight - subjectImg.height + 1), 32, 1);
cvMatchTemplate(colorImg.getCvImage(), subjectImg.getCvImage(), result, CV_TM_SQDIFF);
double minVal, maxVal;
CvPoint minLoc, maxLoc;
cvMinMaxLoc(result, &minVal, &maxVal, &minLoc, &maxLoc, 0);
subjectLocation.x = minLoc.x;
subjectLocation.y = minLoc.y;
}
}
}
void testApp::mouseReleased(int x, int y, int button)
{
//End tracking and normalize subject frame
if(subjectFrame.width < 0)
{
subjectFrame.x += subjectFrame.width;
subjectFrame.width *= -1;
}
if(subjectFrame.height < 0)
{
subjectFrame.y += subjectFrame.height;
subjectFrame.height *= -1;
}
isSelectingTrackingRegion = false;
subjectLocation.x = subjectFrame.x;
subjectLocation.y = subjectFrame.y;
//Copy selected portion of the image to the subject image;
subjectImg.allocate(subjectFrame.width, subjectFrame.height);
colorImg.setROI(subjectFrame);
subjectImg = colorImg;
colorImg.resetROI();
subjectIsDefined = true;
}

Answer 1

You have to use cvReleaseImage() for all the image Iplmage which you have created in your program. You have to be careful to find the correct place to write cvReleaseImage(image name) because once you release the memory related to an image, you can not access that image again.
Suggestion: Don't use IplImage. That's the old c api of OpenCV. Now a days, you can use Mat images which can manage the memory themself. The advantage of using Mat is that you don't need to think about freeing the memory.