opencv c++ 霍夫圆检测

1、原理

        a)对某点(x_{0},y_{0}),以其为圆心的圆为无数(一圈圈的圆),将其从x-y平面坐标系上转换到r-θ极坐标系上后,则变成了以r、θ为自变量,x_{0},y_{0}为固定值,x、y为因变量的式子:

        (x-x_{0})^{2} + (y - y_{0})^{2} = r^{2}

       b)其余点作同样操作,可以得到,当半径r为某值r_{0}时,使得三个圆同时交于1点,从而获取这些点构成的圆的圆心,半径(x_{0},y_{0},r)

        opencv c++ 霍夫圆检测_第1张图片

         圆的参数方程:

        x = x_{0}+ r *cos{\Theta}

        y= y_{0}+ r *sin{\Theta}

        注:在实际实现时,会设定一个固定的半径r来进行检测(因为r的范围太大了)。

2、API

void cv::HoughCircles	(	InputArray 	image,
                            OutputArray 	circles,
                            int 	method,
                            double 	dp,
                            double 	minDist,
                            double 	param1 = 100,
                            double 	param2 = 100,
                            int 	minRadius = 0,
                            int 	maxRadius = 0 
                            )		

image ——输入的灰度图像。
circles——输出,数据类型为vector (x,y,radius) or (x,y,radius,votes) .
method ——检测方法opencv c++ 霍夫圆检测_第2张图片
dp ——累加器分辨率与图像分辨率的反比. For example, if dp=1 , the accumulator has the same resolution as the input image. If dp=2 , the accumulator has half as big width and height. For HOUGH_GRADIENT_ALT the recommended value is dp=1.5, unless some small very circles need to be detected.
minDist ——两个被检测圆的圆心的最小距离,即在这个距离范围内,不会出现第二个被检测出的圆。 If the parameter is too small, multiple neighbor circles may be falsely detected in addition to a true one. If it is too large, some circles may be missed.
param1 ——Canny边缘检测的高阈值。
param2 ——累计阈值,当相交于同一点的圆的个数大于它时,才记录这个被识别到的圆。
minRadius——圆的最小半径
maxRadius ——圆的最大半径。If <= 0, uses the maximum image dimension. If < 0, HOUGH_GRADIENT returns centers without finding the radius. HOUGH_GRADIENT_ALT always computes circle radiuses.

 

3、代码:

        说明:HoughCircles    会基于canny自动二值化图,因此输入灰度图即可,但由于该算法对图像噪点敏感,必须在调用前对灰度图进行降噪处理

void QuickDemo::hough_circle(Mat& image)
{
	//霍夫圆检测会基于canny自动二值化图,因此输入灰度图即可,但在传入之前,需要对图像进行降噪。
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	GaussianBlur(gray, gray, Size(9, 9), 2, 2);
	namedWindow("hough gray", WINDOW_FREERATIO);
	imshow("hough gray", gray);

	vector circles;
	double mindist = 2;
	double min_r = 10;
	double max_r = 200;
	HoughCircles(gray, circles, HOUGH_GRADIENT, 1.5, mindist, 100, 100, min_r, max_r);

	for (size_t i = 0; i < circles.size(); ++i) {
		circle(image, Point(circles[i][0], circles[i][1]), circles[i][2], Scalar(0, 255, 0), 3, 8);
		circle(image, Point(circles[i][0], circles[i][1]), 10, Scalar(250, 0, 0), -1, 8);
	}
	namedWindow("hough circle", WINDOW_FREERATIO);
	imshow("hough circle", image);

}

opencv c++ 霍夫圆检测_第3张图片

 

你可能感兴趣的:(opencv,c++,opencv,人工智能,计算机视觉)