算法是基于A Fast Parallel Algorithm for Thinning Digital Patterns论文

https://blog.csdn.net/keneyr/article/details/88944563

简单的解释算法：

https://blog.csdn.net/xukaiwen_2016/article/details/53135866

#include <opencv2/opencv.hpp>  
#include <opencv2/core/core.hpp>  
#include <iostream>  
#include <vector>  
using namespace cv;
using namespace std;

/**
* @brief 对输入图像进行细化,骨骼化
* @param src为输入图像,用cvThreshold函数处理过的8位灰度图像格式，元素中只有0与1,1代表有元素，0代表为空白
* @param maxIterations限制迭代次数，如果不进行限制，默认为-1，代表不限制迭代次数，直到获得最终结果
* @return 为对src细化后的输出图像,格式与src格式相同，元素中只有0与1,1代表有元素，0代表为空白
*/
cv::Mat thinImage(const cv::Mat & src, const int maxIterations = -1)
{
	assert(src.type() == CV_8UC1);
	cv::Mat dst;
	int width = src.cols;
	int height = src.rows;
	src.copyTo(dst);
	int count = 0;  //记录迭代次数  
	while (true)
	{
		count++;
		if (maxIterations != -1 && count > maxIterations) //限制次数并且迭代次数到达  
			break;
		std::vector<uchar *> mFlag; //用于标记需要删除的点  
									//对点标记  
		for (int i = 0; i < height; ++i)
		{
			uchar * p = dst.ptr<uchar>(i);
			for (int j = 0; j < width; ++j)
			{
				//如果满足四个条件，进行标记  
				//  p9 p2 p3  
				//  p8 p1 p4  
				//  p7 p6 p5  
				uchar p1 = p[j];
				if (p1 != 1) continue;
				uchar p4 = (j == width - 1) ? 0 : *(p + j + 1);
				uchar p8 = (j == 0) ? 0 : *(p + j - 1);
				uchar p2 = (i == 0) ? 0 : *(p - dst.step + j);
				uchar p3 = (i == 0 || j == width - 1) ? 0 : *(p - dst.step + j + 1);
				uchar p9 = (i == 0 || j == 0) ? 0 : *(p - dst.step + j - 1);
				uchar p6 = (i == height - 1) ? 0 : *(p + dst.step + j);
				uchar p5 = (i == height - 1 || j == width - 1) ? 0 : *(p + dst.step + j + 1);
				uchar p7 = (i == height - 1 || j == 0) ? 0 : *(p + dst.step + j - 1);
				if ((p2 + p3 + p4 + p5 + p6 + p7 + p8 + p9) >= 2 && (p2 + p3 + p4 + p5 + p6 + p7 + p8 + p9) <= 6)
				{
					int ap = 0;
					if (p2 == 0 && p3 == 1) ++ap;
					if (p3 == 0 && p4 == 1) ++ap;
					if (p4 == 0 && p5 == 1) ++ap;
					if (p5 == 0 && p6 == 1) ++ap;
					if (p6 == 0 && p7 == 1) ++ap;
					if (p7 == 0 && p8 == 1) ++ap;
					if (p8 == 0 && p9 == 1) ++ap;
					if (p9 == 0 && p2 == 1) ++ap;

					if (ap == 1 && p2 * p4 * p6 == 0 && p4 * p6 * p8 == 0)
					{
						//标记  
						mFlag.push_back(p + j);
					}
				}
			}
		}

		//将标记的点删除  
		for (std::vector<uchar *>::iterator i = mFlag.begin(); i != mFlag.end(); ++i)
		{
			**i = 0;
		}

		//直到没有点满足，算法结束  
		if (mFlag.empty())
		{
			break;
		}
		else
		{
			mFlag.clear();//将mFlag清空  
		}
	}

	return dst;
}

/**
* @brief 对骨骼化图数据进行过滤，实现两个点之间至少隔一个空白像素
* @param thinSrc为输入的骨骼化图像,8位灰度图像格式，元素中只有0与1,1代表有元素，0代表为空白
*/
void filterOver(cv::Mat thinSrc)
{
	assert(thinSrc.type() == CV_8UC1);
	int width = thinSrc.cols;
	int height = thinSrc.rows;
	for (int i = 0; i < height; ++i)
	{
		uchar * p = thinSrc.ptr<uchar>(i);
		for (int j = 0; j < width; ++j)
		{
			// 实现两个点之间至少隔一个像素
			//  p9 p2 p3  
			//  p8 p1 p4  
			//  p7 p6 p5  
			uchar p1 = p[j];
			if (p1 != 1) {
				continue;
			}
			uchar p4 = (j == width - 1) ? 0 : *(p + j + 1);
			uchar p8 = (j == 0) ? 0 : *(p + j - 1);
			uchar p2 = (i == 0) ? 0 : *(p - thinSrc.step + j);
			uchar p3 = (i == 0 || j == width - 1) ? 0 : *(p - thinSrc.step + j + 1);
			uchar p9 = (i == 0 || j == 0) ? 0 : *(p - thinSrc.step + j - 1);
			uchar p6 = (i == height - 1) ? 0 : *(p + thinSrc.step + j);
			uchar p5 = (i == height - 1 || j == width - 1) ? 0 : *(p + thinSrc.step + j + 1);
			uchar p7 = (i == height - 1 || j == 0) ? 0 : *(p + thinSrc.step + j - 1);
			if (p2 + p3 + p8 + p9 >= 1)
			{
				p[j] = 0;
			}
		}

	}
}

/**
* @brief 从过滤后的骨骼化图像中寻找端点和交叉点
* @param thinSrc为输入的过滤后骨骼化图像,8位灰度图像格式，元素中只有0与1,1代表有元素，0代表为空白
* @param raudis卷积半径，以当前像素点位圆心，在圆范围内判断点是否为端点或交叉点
* @param thresholdMax交叉点阈值，大于这个值为交叉点
* @param thresholdMin端点阈值，小于这个值为端点
* @return 为对src细化后的输出图像,格式与src格式相同，元素中只有0与1,1代表有元素，0代表为空白
*/
std::vector<cv::Point> getPoints(const cv::Mat &thinSrc, unsigned int raudis = 4, unsigned int thresholdMax = 6, unsigned int thresholdMin = 4)
{
	assert(thinSrc.type() == CV_8UC1);
	int width = thinSrc.cols;
	int height = thinSrc.rows;
	cv::Mat tmp;
	thinSrc.copyTo(tmp);
	std::vector<cv::Point> points;
	cout << tmp.step << endl;
	for (int i = 0; i < height; ++i)
	{
		for (int j = 0; j < width; ++j)
		{
			if (*(tmp.data + tmp.step * i + j) == 0)
			{
				continue;
			}
			int count = 0;
			for (int k = i - raudis; k < i + raudis + 1; k++)
			{
				for (int l = j - raudis; l < j + raudis + 1; l++)
				{
					if (k < 0 || l < 0 || k>height - 1 || l>width - 1)
					{
						continue;

					}
					else if (*(tmp.data + tmp.step * k + l) == 1)
					{
						count++;
					}
				}
			}

			if (count > thresholdMax || count<thresholdMin)
			{
				Point point(j, i);
				points.push_back(point);
			}
		}
	}
	return points;
}


int main(int argc, char*argv[])
{
	cv::Mat src;
	//获取图像  
	src = cv::imread("binary.jpg", cv::IMREAD_GRAYSCALE);

	//将原图像转换为二值图像  
	cv::threshold(src, src, 20, 1, cv::THRESH_BINARY);
	//图像细化，骨骼化  
	cv::Mat dst = thinImage(src);
	cv::Mat result = cv::Mat::zeros(dst.size(), CV_8UC3);
	//过滤细化后的图像
	filterOver(dst);
	//查找端点和交叉点  
	std::vector<cv::Point> points = getPoints(dst, 4,6, 4);
	//二值图转化成灰度图，并绘制找到的点
	dst = dst * 255;
	cv::cvtColor(dst, result, cv::COLOR_GRAY2BGR);

	vector<cv::Point>::iterator it = points.begin();
	for (; it != points.end(); it++)
	{
		//circle(result, *it, 2, cv::Scalar(255,0,0), -1);
	}

	imwrite("dst.jpg", result);
	//显示图像  
	//cv::namedWindow("dst1", CV_WINDOW_AUTOSIZE);
	//cv::imshow("dst1", result);
	//cv::waitKey(0);
}