Opencv notes 17 template matching

  One 、 Template matching

Template matching （TemplateMatching） Is to find and template images in an image （template） The most similar area , This method has simple principle and fast calculation speed , It can be applied to target recognition , Target tracking and other fields .

Two 、 principle

1、cv::TM_SQDIFF： This method uses square difference for matching , Therefore, the best matching result is 0 It's about , The larger the value, the worse the matching result .

2、cv::TM_SQDIFF_NORMED： The method uses the normalized square difference for matching , The best match is also when the result is 0 It's about .

3、cv::TM_CCORR： Correlation matching method , This method uses the convolution results of source image and template image to match , therefore , The best match position is at the maximum value , The smaller the value, the worse the matching result .

4、cv::TM_CCORR_NORMED： Normalized correlation matching method , Similar to the correlation matching method , The best match position is also at the maximum value .

5、cv::TM_CCOEFF： Correlation coefficient matching method , This method uses the difference between the source image and its mean 、 The correlation between the difference between the template and its mean is matched , The best match result is when the value is equal to 1 It's about , The worst match results in a value equal to -1 It's about , The value is equal to 0 Direct means that the two are not related .

6、cv::TM_CCOEFF_NORMED： Normalized correlation coefficient matching method , A positive value indicates a good match , A negative value indicates a poor match , Also, the greater the value , The matching effect is good .

【OpenCV3】 Template matching ——cv::matchTemplate() Detailed explanation _guduruyu The column -CSDN Blog _matchtemplate function

3、 ... and 、API

MatchTemplate(
 InputArrray image,  // The source image 
InputArray templ,   // Template image 
OutputArray result,// Output results    It has to be a single channel 32 Bit floating point , Suppose the source image WxH, Template image   w*h, The result must be W-w+1,H-h+1 Size 
int method   //  Match method , The meaning of the results produced by different methods may be different , The larger the returned value, the better the matching degree , And the smaller the value returned by some methods, the better the matching degree 

 About parameters  method：

CV_TM_SQDIFF Square error matching ： This method uses square difference to match ; The best match value is 0; Match the worse , The larger the match value .
CV_TM_CCORR Correlation matching ： The method adopts multiplication operation ; The larger the number, the better the match .
CV_TM_CCOEFF Correlation coefficient matching method ：1 A perfect match ;-1 Represents the worst match .
CV_TM_SQDIFF_NORMED Normalized square error matching method 
CV_TM_CCORR_NORMED Normalized correlation matching method 
CV_TM_CCOEFF_NORMED Normalized correlation coefficient matching method 
);

Four 、 Code shows

 

#if  1 //  Template matching    There are six ways 
Mat src, tempalte, result;
int matchMethod;
int maxTrackbarType = 5;


void on_matching(int, void*)
{
	Mat srcImage;
	src.copyTo(srcImage);
	int result_cols = src.cols - tempalte.cols + 1;
	int result_rows = src.rows - tempalte.rows + 1;


	cout << "---- result_rows----: " << result_rows << endl;
	cout << "---- result_cols----: " << result_cols << endl;



	result.create(result_cols, result_rows, CV_32FC1);

	//  Begin to match 
	matchTemplate(src, tempalte, result, matchMethod);


	normalize(result, result, 0, 2, NORM_MINMAX, -1, Mat());

	double minValue, maxValue;

	Point minLocation, maxLocation, matchLocation;
	minMaxLoc(result, &minValue, &maxValue, &minLocation, &maxLocation);

	if (matchMethod == TM_SQDIFF || matchMethod == CV_TM_SQDIFF_NORMED)
	{
		matchLocation = minLocation;
	}
	else
	{
		matchLocation = maxLocation;
	}

	rectangle(srcImage, matchLocation, Point(matchLocation.x + tempalte.cols, matchLocation.y + tempalte.rows), Scalar(0, 0, 255), 2, 8, 0);
	rectangle(result, matchLocation, Point(matchLocation.x + tempalte.cols, matchLocation.y + tempalte.rows), Scalar(0, 0, 255), 2, 8, 0);

	imshow(" The original picture ", srcImage);
	imshow(" design sketch ", result);

}

int main()
{

	src = imread("C:\\Users\\19473\\Desktop\\opencv_images\\609.png");
	if (!src.data)
	{
		cout << " Failed to read the original graph " << endl;
		return -1;
	}
	cout << " Original target image   " << src.rows << "    ,    " << src.cols << endl;
	tempalte = imread("C:\\Users\\19473\\Desktop\\opencv_images\\610.png");
	if (!tempalte.data)
	{
		cout << " Template map reading failed " << endl;
		return -1;
	}
	cout << " Target target image   " << tempalte.rows << "    ,    " << tempalte.cols << endl;


	imshow(" Templates ", tempalte);
	namedWindow(" The original picture ", CV_WINDOW_AUTOSIZE);
	namedWindow(" design sketch ", CV_WINDOW_AUTOSIZE);
	createTrackbar(" Method ", " The original picture ", &matchMethod, maxTrackbarType, on_matching);

	on_matching(0, NULL);


	waitKey(0);

	return 0;
}
#endif