图像处理——（源）最大熵阈值分割（Max_Entropy）函数编程实现 


  1 #include <iostream>
  2 #include <opencv2/core.hpp>
  3 #include <opencv2/highgui.hpp>
  4 #include <opencv2/imgproc.hpp>
  5 
  6  
  7 int Max_Entropy(cv::Mat& src, cv::Mat& dst, int thresh, int p){
  8     const int Grayscale = 256;
  9     int Graynum[Grayscale] = { 0 };
 10     int r = src.rows;
 11     int c = src.cols;
 12     for (int i = 0; i < r; ++i){
 13         const uchar* ptr = src.ptr<uchar>(i);   
 14         for (int j = 0; j < c; ++j){   
 15             if (ptr[j] == 0)                //排除掉黑色的像素点
 16                 continue;
 17             Graynum[ptr[j]]++;
 18         }
 19     }
 20  
 21     float probability = 0.0; //概率
 22     float max_Entropy = 0.0; //最大熵
 23     int totalpix = r*c;
 24     for (int i = 0; i < Grayscale; ++i){
 25  
 26         float HO = 0.0; //前景熵
 27         float HB = 0.0; //背景熵
 28  
 29         //计算前景像素数
 30         int frontpix = 0;
 31         for (int j = 0; j < i; ++j){
 32             frontpix += Graynum[j];
 33         }
 34         //计算前景熵
 35         for (int j = 0; j < i; ++j){
 36             if (Graynum[j] != 0){
 37                 probability = (float)Graynum[j] / frontpix;
 38                 HO = HO + probability*log(1/probability);
 39             }
 40         }
 41  
 42         //计算背景熵
 43         for (int k = i; k < Grayscale; ++k){
 44             if (Graynum[k] != 0){
 45                 probability = (float)Graynum[k] / (totalpix - frontpix);
 46                 HB = HB + probability*log(1/probability);
 47             }
 48         }
 49  
 50         //计算最大熵
 51         if(HO + HB > max_Entropy){
 52             max_Entropy = HO + HB;
 53             thresh = i + p;
 54         }
 55     }
 56  
 57     //阈值处理
 58     src.copyTo(dst);
 59     for (int i = 0; i < r; ++i){
 60         uchar* ptr = dst.ptr<uchar>(i);
 61         for (int j = 0; j < c; ++j){
 62             if (ptr[j]> thresh)
 63                 ptr[j] = 255;
 64             else
 65                 ptr[j] = 0;
 66         }
 67     }
 68     return thresh;
 69 }
 70  
 71  
 72 int main(){
 73     cv::Mat src = cv::imread("E://lena.jpg");
 74     if (src.empty()){
 75         return -1;
 76     }
 77     if (src.channels() > 1)
 78         cv::cvtColor(src, src, CV_RGB2GRAY);
 79  
 80     cv::Mat dst, dst2;
 81     int thresh = 0;
 82     double t2 = (double)cv::getTickCount();
 83     thresh = Max_Entropy(src, dst, thresh,10); //Max_Entropy
 84     std::cout << "Mythresh=" << thresh << std::endl;
 85     t2 = (double)cv::getTickCount() - t2;
 86     double time2 = (t2 *1000.) / ((double)cv::getTickFrequency());
 87     std::cout << "my_process=" << time2 << " ms. " << std::endl << std::endl;
 88  
 89     double  Otsu = 0;
 90     Otsu = cv::threshold(src, dst2, Otsu, 255, CV_THRESH_OTSU + CV_THRESH_BINARY);
 91     std::cout << "Otsuthresh=" << Otsu << std::endl;
 92     
 93  
 94     cv::namedWindow("src", CV_WINDOW_NORMAL);
 95     cv::imshow("src", src);
 96     cv::namedWindow("dst", CV_WINDOW_NORMAL);
 97     cv::imshow("dst", dst);
 98     cv::namedWindow("dst2", CV_WINDOW_NORMAL);
 99     cv::imshow("dst2", dst2);
100     
101     cv::waitKey(0);
102 }

 

posted on 2019-07-20 21:42  AI大道理  阅读(1212)  评论(0)  编辑  收藏  举报

导航