opencv_28 模板匹配

1.概念
建立一个模板子区在源图像中从上到下从左到右寻找与模板子区最相似的像素块

2.六种相似性算法
(推荐使用归一化方法，结果在0-1之间好判断)

• 计算平方不同TM_SQDIFF 0
  得到的值越小，则说明越相似

• 计算归一化平方不同 TM_SQDIFF_NORMED 1

• 计算相关性TM_CCORR 2
  得到的值越大，则说明越相似

• 计算归一化相关性TM_CCORR_NORMED 3

• 计算相关系数 TM_CCOEFF 4
  得到的值越大，则说明越相似

• 计算归一化相关系数TM_CCOEFF_NORMED 5

3.模板匹配 API

matchTemplate{
input image1;//源图像8位 or32位 浮点数图像
input image2;//模板图像 与源图像类型一致
output result;//输出单通道32位浮点数图像，源图像W*H,模板图像w*h,则结果为W-w+1 * H-h+1
input method;//匹配方法
InputArray mask=noArray()//(optional)
]

• 匹配结果

src.copyTo(img_display);
	dst.create(result_col, result_rows, CV_32FC1);

• 模板匹配

matchTemplate(src, temp, dst, match_method, Mat());
	normalize(dst, dst, 0, 1, NORM_MINMAX, -1, Mat());

• 在图像中找到匹配的位置
  1.找到匹配到的模板左上角点坐标

Point minLoc;
	Point maxLoc;
	double min, max;
	Point match_Loc;
	minMaxLoc(dst, &min, &max, &minLoc, &maxLoc, Mat());//在末班输出图像找到最大
	if (match_method == TM_SQDIFF || TM_SQDIFF_NORMED)
		match_Loc = minLoc;
	else {
		match_Loc = maxLoc;
	}

• 绘制ROI

rectangle(src, Rect(match_Loc.x, match_Loc.y, temp.cols, temp.rows), Scalar(0, 0, 255), 2, LINE_AA);
	rectangle(img_display, Rect(match_Loc.x, match_Loc.y, temp.cols, temp.rows), Scalar(0, 0, 255), 2, LINE_AA);

Mat src, dst, temp, result;
int match_method = TM_SQDIFF;
int max_track = 5;
const char* INPUT_T = "input image";
const char* OUTPUT_T = "result image";
const char* match_t = "template match-demo";
void Match_Demo(int, void*);

int main(int argc, char** argv) {
	src = imread("C:/Users/Administrator/Pictures/Saved Pictures/timg6ZAGKYHR.jpg");//待检测图像 
	temp = imread("C:/Users/Administrator/Pictures/Saved Pictures/template.jpg");//模板图像
	if (!src.data || !temp.data) {
		cout << "could not load image..." << endl;
		return -1;
	}
	namedWindow(INPUT_T, WINDOW_AUTOSIZE);
	namedWindow(OUTPUT_T, WINDOW_AUTOSIZE);
	//namedWindow(match_t, WINDOW_AUTOSIZE);
	//imshow(INPUT_T, src);
	imshow(match_t, temp);

	const char* trackbar_title = "Match Alogo Type";
	createTrackbar(trackbar_title, OUTPUT_T, &match_method, 5, Match_Demo);
	Match_Demo(0, 0);

	waitKey(0);
	return 0;
}

void Match_Demo(int, void*) {
	int result_rows = src.rows - temp.rows + 1;
	int result_col = src.cols - temp.cols + 1;
	Mat img_display;
	src.copyTo(img_display);
	dst.create(result_col, result_rows, CV_32FC1);
	matchTemplate(src, temp, dst, match_method, Mat());
	normalize(dst, dst, 0, 1, NORM_MINMAX, -1, Mat());
	Point minLoc;
	Point maxLoc;
	double min, max;
	Point match_Loc;
	minMaxLoc(dst, &min, &max, &minLoc, &maxLoc, Mat());//在末班输出图像找到最大
	if (match_method == TM_SQDIFF || TM_SQDIFF_NORMED)
		match_Loc = minLoc;
	else {
		match_Loc = maxLoc;
	}
	rectangle(src, Rect(match_Loc.x, match_Loc.y, temp.cols, temp.rows), Scalar(0, 0, 255), 2, LINE_AA);
	rectangle(img_display, Rect(match_Loc.x, match_Loc.y, temp.cols, temp.rows), Scalar(0, 0, 255), 2, LINE_AA);
	imshow(INPUT_T,src);
	imshow(OUTPUT_T, dst);
}