【OpenCV入門之十二】看起來一樣的圖像竟然存在這麼大的差別!

對輸入的兩張圖像計算得到直方圖H1與H2，歸一化到相同的尺度空間然後可以通過計算H1與H2的之間的距離得到兩個直方圖的相似程度進而比較圖像本身的相似程度。Opencv提供的比較方法有四種：

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

using namespace std;
using namespace cv;

string convertToString(double d);

int main(int argc, char** argv) {
 
 Mat base, test1, test2;
 Mat hsvbase, hsvtest1, hsvtest2;
 
 base = imread("E:/Experiment/OpenCV/Pictures/lenanoise.jpg");
 if (!base.data) {
   printf("could not load image...\n");
   return -1;
 }
 test1 = imread("E:/Experiment/OpenCV/Pictures/lena.jpg");
 test2 = imread("E:/Experiment/OpenCV/Pictures/lena1.jpg");
 
 cvtColor(base, hsvbase, CV_BGR2HSV);
 cvtColor(test1, hsvtest1, CV_BGR2HSV);
 cvtColor(test2, hsvtest2, CV_BGR2HSV);

 
 int h_bins = 50; int s_bins = 60;    
 int histSize[] = { h_bins, s_bins };
 
 float h_ranges[] = { 0, 180 };    
 float s_ranges[] = { 0, 256 };
 const float* ranges[] = { h_ranges, s_ranges };
 
 int channels[] = { 0, 1 };
   
 
 
 MatND hist_base;
 MatND hist_test1;
 MatND hist_test2;

 
 calcHist(&hsvbase, 1,  channels, Mat(), hist_base, 2, histSize, ranges, true, false);
 normalize(hist_base, hist_base, 0, 1, NORM_MINMAX, -1, Mat());

 calcHist(&hsvtest1, 1, channels, Mat(), hist_test1, 2, histSize, ranges, true, false);
 normalize(hist_test1, hist_test1, 0, 1, NORM_MINMAX, -1, Mat());

 calcHist(&hsvtest2, 1, channels, Mat(), hist_test2, 2, histSize, ranges, true, false);
 normalize(hist_test2, hist_test2, 0, 1, NORM_MINMAX, -1, Mat());
 

 
 
 double basebase = compareHist(hist_base, hist_base, CV_COMP_CORREL);
 double basetest1 = compareHist(hist_base, hist_test1, CV_COMP_CORREL);
 double basetest2 = compareHist(hist_base, hist_test2, CV_COMP_CORREL);
 double tes1test2 = compareHist(hist_test1, hist_test2, CV_COMP_CORREL);
 printf("test1 compare with test2 correlation value :%f", tes1test2);

 Mat test12;
 test2.copyTo(test12);
 
 putText(base, convertToString(basebase), Point(50, 50), CV_FONT_HERSHEY_COMPLEX, 1, Scalar(0, 0, 255), 2, LINE_AA);
 putText(test1, convertToString(basetest1), Point(50, 50), CV_FONT_HERSHEY_COMPLEX, 1, Scalar(0, 0, 255), 2, LINE_AA);
 putText(test2, convertToString(basetest2), Point(50, 50), CV_FONT_HERSHEY_COMPLEX, 1, Scalar(0, 0, 255), 2, LINE_AA);
 putText(test12, convertToString(tes1test2), Point(50, 50), CV_FONT_HERSHEY_COMPLEX, 1, Scalar(0, 0, 255), 2, LINE_AA);
 
 namedWindow("base", CV_WINDOW_AUTOSIZE);
 namedWindow("test1", CV_WINDOW_AUTOSIZE);
 namedWindow("test2", CV_WINDOW_AUTOSIZE);

 imshow("base", base);
 imshow("test1", test1);
 imshow("test2", test2);
 imshow("test12", test12);

 waitKey(0);
 return 0;
}

string convertToString(double d) {
 ostringstream os;
 if (os << d)  
   return os.str();
 return "invalid conversion";
}