opencv实现导向滤波



*******************************************************************//
*****************调用opencv实现导向滤波****************************//
*******************************************************************//
#include <opencv2/core/core.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/imgproc/types_c.h>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/highgui/highgui_c.h>
#include <iostream>

using namespace cv;
using namespace std;

//导向滤波器 
Mat guidedfilter(Mat &srcImage, Mat &srcClone, int r, double eps);

int main()
{
  Mat srcImage = imread("F:\\wangjiao\\picture\\lena.jpg");
  if (srcImage.empty())
  {
   cout << "读入图片错误！" << endl;
   system("pause");
   return -1;
  }
  //进行通道分离
  vector<Mat>vSrcImage, vResultImage;
  split(srcImage, vSrcImage);
  Mat resultMat;
  for (int i = 0; i < 3; i++)
  {
   //分通道转换成浮点型数据
   Mat tempImage;
   vSrcImage[i].convertTo(tempImage, CV_64FC1, 1.0 / 255.0);
 //  imwrite("F:\\wangjiao\\20180420\\src_Bilateral_yuv.jpg", tempImage*255);
   Mat p = tempImage.clone();
   //分别进行导向滤波
   Mat resultImage = guidedfilter(tempImage, p, 5, 0.01);
   resultImage.convertTo(resultImage, CV_8UC1, 255.0);
   vResultImage.push_back(resultImage);
  }
  //通道结果合并
  merge(vResultImage, resultMat);
  imshow("原图像", srcImage);
  imshow("导向滤波后图像", resultMat);
  waitKey(0);
  return 0;
}

Mat guidedfilter(Mat &srcImage, Mat &srcClone, int r, double eps)
{
 //转换源图像信息 
 srcImage.convertTo(srcImage, CV_64FC1);
 srcClone.convertTo(srcClone, CV_64FC1);
 int NumRows = srcImage.rows;
 int NumCols = srcImage.cols;
 Mat boxResult;

 //下面按照步骤进行导向滤波操作 
 / 
 //步骤一：计算均值 
 boxFilter(Mat::ones(NumRows, NumCols, srcImage.type()),
  boxResult, CV_64FC1, Size(r, r));
 //生成导向均值mean_I 
 Mat mean_I;
 boxFilter(srcImage, mean_I, CV_64FC1, Size(r, r));//mean_I为uk
 //生成原始均值mean_P 
 Mat mean_P;
 boxFilter(srcClone, mean_P, CV_64FC1, Size(r, r));//mean_P为pk
 //生成互相关均值mean_IP 
 Mat mean_IP;
 boxFilter(srcImage.mul(srcClone), mean_IP,
  CV_64FC1, Size(r, r));//ak表达式中的分子的第一项
 Mat cov_IP = mean_IP - mean_I.mul(mean_P);
 //生成自相关均值mean_II 
 Mat mean_II;
 //应用盒滤波计算相关均值 
 boxFilter(srcImage.mul(srcImage), mean_II, CV_64FC1, Size(r, r));
 //步骤二：计算相关系数 
 Mat var_I = mean_II - mean_I.mul(mean_I);//
 Mat var_IP = mean_IP - mean_I.mul(mean_P);
 //步骤三:计算参数系数a,b 
 Mat a = cov_IP / (var_I + eps);
 Mat b = mean_P - a.mul(mean_I);
 //步骤四：计算系数a,b的均值 
 Mat mean_a;
 boxFilter(a, mean_a, CV_64FC1, Size(r, r));
 mean_a = mean_a / boxResult;
 Mat mean_b;
 boxFilter(b, mean_b, CV_64FC1, Size(r, r));
 mean_b = mean_b / boxResult;
 //步骤五：生成输出矩阵 
 Mat resultMat = mean_a.mul(srcImage) + mean_b;
 imwrite("F:\\wangjiao\\20180420\\src_Bilateral_r.jpg", resultMat*255);
 return resultMat;
}