【opencv】轮廓高斯滤波平滑

一个小测试的题目：

在图像上点选，找到与点选处相邻的颜色相近的点集合，对该点集合提取轮廓，对该点集合轮廓平滑处理，显示处理结果。

#include <opencv2/opencv.hpp>
#include <iostream>

using namespace std;
using namespace cv;


//************************************
// Method:    findRegion 漫水填充
// FullName:  findRegion
// Access:    public 
// Returns:   vector<Point>
// Qualifier:
// Parameter: Mat img
// Parameter: Point pos
// Parameter: int LowDifference
// Parameter: int UpDifference
//************************************
vector<Point> findRegion(Mat img, Point pos, int LowDifference, int UpDifference)
{
	Mat image = img.clone();
	Mat imgBack = img.clone();
	Rect ccomp;
	Scalar pixel = image.at<Vec3b>(pos);
	Scalar pixel2 = Scalar(255 - pixel[0], 255 - pixel[1], 255 - pixel[2], pixel[3]);
	floodFill(image, pos, pixel2, &ccomp, Scalar(LowDifference, LowDifference, LowDifference),
		Scalar(UpDifference, UpDifference, UpDifference));

	Mat diff;
	absdiff(image, imgBack, diff);

	//统计所有非零像素
	vector<Point> pt;
	for (int i = 0; i < diff.rows; i++)
	{
		for (int j = 0; j < diff.cols; j++)
		{
			Point newPos(j, i);
			Scalar pixel3 = diff.at<Vec3b>(newPos);
			if (pixel3[0] != 0 || pixel3[1] != 0 || pixel3[2] != 0)
			{
				pt.push_back(newPos);
			}
		}
	}

	return pt;
}

//************************************
// Method:    findPerimeter 从点集合中寻找轮廓点
// FullName:  findPerimeter
// Access:    public 
// Returns:   vector<Point>
// Qualifier:
// Parameter: vector<Point> pt
// Parameter: Size size
//************************************
vector<Point> findPerimeter(vector<Point> pt, Size size)
{
	Mat imgGray(size, CV_8UC1, Scalar(0));
	for (int i = 0; i < pt.size(); i++)
	{
		imgGray.at<uchar>(pt[i]) = 255;
	}

	std::vector<std::vector<cv::Point>> contours;
	//获取轮廓不包括轮廓内的轮廓    
	cv::findContours(imgGray.clone(), contours, CV_RETR_EXTERNAL, CV_CHAIN_APPROX_NONE);

	return contours[0];
}

//************************************
// Method:    displayImage 显示图像
// FullName:  displayImage
// Access:    public 
// Returns:   void
// Qualifier:
// Parameter: Mat img
// Parameter: vector<Point> contours 轮廓点
// Parameter: Point pos
//************************************
void displayImage(Mat img, vector<Point> contours, Point pos)
{
	Mat imgShow = img.clone();
	for (int i = 0; i < contours.size(); i++)
	{
		imgShow.at<Vec3b>(contours[i])[0] = 0;
		imgShow.at<Vec3b>(contours[i])[1] = 0;
		imgShow.at<Vec3b>(contours[i])[2] = 0;
	}

	circle(imgShow, pos, 3, Scalar(0, 0, 0), 1, 8, 0);//画用户选择的点

	imshow("img", imgShow);
	waitKey(0);
}


//************************************
// Method:    findSmoothPeimeter 高斯滤波轮廓点平滑
// FullName:  findSmoothPeimeter
// Access:    public 
// Returns:   void
// Qualifier:
// Parameter: Mat img 原图
// Parameter: vector<Point> pt 轮廓点集合
//************************************
void findSmoothPeimeter(Mat img, vector<Point> pt)
{
	vector<Point> contours = findPerimeter(pt, img.size());

	Mat im;
	cvtColor(img, im, CV_BGR2GRAY);

	Mat cont = ~im;
	Mat original = Mat::zeros(im.rows, im.cols, CV_8UC3);
	Mat smoothed = img.clone();

	// contour smoothing parameters for gaussian filter
	int filterRadius = 10;
	int filterSize = 2 * filterRadius + 1;
	double sigma = 10;

	size_t len = contours.size() + 2 * filterRadius;
	size_t idx = (contours.size() - filterRadius);
	vector<float> x, y;
	for (size_t i = 0; i < len; i++)
	{
		x.push_back(contours[(idx + i) % contours.size()].x);
		y.push_back(contours[(idx + i) % contours.size()].y);
	}
	// filter 1-D signals
	vector<float> xFilt, yFilt;
	GaussianBlur(x, xFilt, Size(filterSize, filterSize), sigma, sigma);
	GaussianBlur(y, yFilt, Size(filterSize, filterSize), sigma, sigma);
	// build smoothed contour
	vector<Point> smoothContours;
	for (size_t i = filterRadius; i < contours.size() + filterRadius; i++)
	{
		smoothContours.push_back(Point(xFilt[i], yFilt[i]));
	}

	Scalar color;

	for (int i = 0; i < smoothContours.size(); i++)
	{
		smoothed.at<Vec3b>(smoothContours[i])[0] = 0;
		smoothed.at<Vec3b>(smoothContours[i])[1] = 0;
		smoothed.at<Vec3b>(smoothContours[i])[2] = 0;
	}

	imshow("smoothed", smoothed);
	waitKey(10);
}

void main()
{
	Mat img = imread("4.jpg", 1);

	vector<Point> pt, contours;
	Point pos(1438, 590);//先列后行
	int para1 = 2;
	int para2 = 2;
	pt = findRegion(img, pos, para1, para2);
	findSmoothPeimeter(img, pt);

	contours = findPerimeter(pt, img.size());//轮廓点集合
	displayImage(img, contours, pos);//显示图像
}

               漫水填充找到的轮廓                                          轮廓滤波平滑