引导滤波 Guided Image Filtering

本文主要介绍导向滤波，引导滤波在滤波的同时，具有保边缘的特效，考虑了空间因素，可以用在图像去雾算法中对透射率t的优化中，能起到很好的效果

 

论文如下：

Guided Image Filtering
Kaiming He  , Jian Sun , and Xiaoou Tang ,
Department of Information Engineering, The Chinese University of Hong Kong
Microsoft Research Asia
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, China

原理如下：

 

代码如下：

#include "afxwin.h"
#include 
#include "cxcore.h"  
#include 
#include
#include   
#include "opencv2/core/core.hpp"    
#include "opencv2/highgui/highgui.hpp"    
#include "opencv2/imgproc/imgproc.hpp"   

void Ctry::OnTryTyr1()  //消息响应函数
{
	int r = 4;
	double eps = 0.01;

	Mat image_src = imread("C:\\Users\\Administrator\\Desktop\\38.jpg", CV_LOAD_IMAGE_COLOR);
	vector bgr_src, bgr_dst;
	split(image_src, bgr_src);//分解每个通道  

	Mat dst_color;
	Mat temp;
	for (int i = 0; i<3; i++)
	{
		Mat I = getimage(bgr_src[i]);
		Mat p = I.clone();
		Mat q = guidedFilter2(I, p, r, eps);	
		bgr_dst.push_back(q);
	}
	merge(bgr_dst, dst_color);

	imwrite("C:\\Users\\Administrator\\Desktop\\result2.jpg", dst_color*255);


	double time2 = 0;
	time2 = (double)getTickCount();
	Mat image_gray(image_src.size(), CV_8UC1);
	cvtColor(image_src, image_gray, CV_BGR2GRAY);
	Mat I = getimage(image_gray);
	Mat p = I.clone();
	
	Mat  Output;
	guidedFilter2(I, p, r, eps).copyTo(Output);
	
	imshow("方法：", Output);
	Mat tt;
	Output.convertTo(tt, -1,255.0);
	imwrite("C:\\Users\\Administrator\\Desktop\\mert.JPG", tt);

	time2 = 1000 * ((double)getTickCount() - time2) / getTickFrequency();
	cout << endl << "Time of guided filter2 for  runs: " << time2 << " milliseconds." << endl;	
	waitKey(0);
}

//convert image depth to CV_64F  
Mat  Ctry::getimage(Mat &a)
{
	int hei = a.rows;
	int wid = a.cols;
	Mat I(hei, wid, CV_64FC1);
	//convert image depth to CV_64F  
	a.convertTo(I, CV_64FC1, 1.0 / 255.0);
	return I;
}

Mat  Ctry::guidedFilter2(cv::Mat I, cv::Mat p, int r, double eps)
{
	/*
	% GUIDEDFILTER   O(1) time implementation of guided filter.
	%
	%   - guidance image: I (should be a gray-scale/single channel image)
	%   - filtering input image: p (should be a gray-scale/single channel image)
	%   - local window radius: r
	%   - regularization parameter: eps
	*/

	cv::Mat _I;
	I.convertTo(_I, CV_64FC1);
	I = _I;

	cv::Mat _p;
	p.convertTo(_p, CV_64FC1);
	p = _p;

	//[hei, wid] = size(I);  
	int hei = I.rows;
	int wid = I.cols;

	//N = boxfilter(ones(hei, wid), r); % the size of each local patch; N=(2r+1)^2 except for boundary pixels.  
	cv::Mat N;
	cv::boxFilter(cv::Mat::ones(hei, wid, I.type()), N, CV_64FC1, cv::Size(r, r));

	//mean_I = boxfilter(I, r) ./ N;  
	cv::Mat mean_I;
	cv::boxFilter(I, mean_I, CV_64FC1, cv::Size(r, r));

	//mean_p = boxfilter(p, r) ./ N;  
	cv::Mat mean_p;
	cv::boxFilter(p, mean_p, CV_64FC1, cv::Size(r, r));

	//mean_Ip = boxfilter(I.*p, r) ./ N;  
	cv::Mat mean_Ip;
	cv::boxFilter(I.mul(p), mean_Ip, CV_64FC1, cv::Size(r, r));

	//cov_Ip = mean_Ip - mean_I .* mean_p; % this is the covariance of (I, p) in each local patch.  
	cv::Mat cov_Ip = mean_Ip - mean_I.mul(mean_p);

	//mean_II = boxfilter(I.*I, r) ./ N;  
	cv::Mat mean_II;
	cv::boxFilter(I.mul(I), mean_II, CV_64FC1, cv::Size(r, r));

	//var_I = mean_II - mean_I .* mean_I;  
	cv::Mat var_I = mean_II - mean_I.mul(mean_I);

	//a = cov_Ip ./ (var_I + eps); % Eqn. (5) in the paper;     
	cv::Mat a = cov_Ip / (var_I + eps);

	//b = mean_p - a .* mean_I; % Eqn. (6) in the paper;  
	cv::Mat b = mean_p - a.mul(mean_I);

	//mean_a = boxfilter(a, r) ./ N;  
	cv::Mat mean_a;
	cv::boxFilter(a, mean_a, CV_64FC1, cv::Size(r, r));
	mean_a = mean_a / N;

	//mean_b = boxfilter(b, r) ./ N;  
	cv::Mat mean_b;
	cv::boxFilter(b, mean_b, CV_64FC1, cv::Size(r, r));
	mean_b = mean_b / N;

	//q = mean_a .* I + mean_b; % Eqn. (8) in the paper;  
	cv::Mat q = mean_a.mul(I) + mean_b;

	return q;
}

效果图：