目标

本文档尝试解答如下问题:

原理

源码

#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include <iostream>
#include <stdio.h>using namespace std;
using namespace cv;/** @函数 main */
int main( int argc, char** argv )
{Mat src_base, hsv_base;Mat src_test1, hsv_test1;Mat src_test2, hsv_test2;Mat hsv_half_down;/// 装载三张背景环境不同的图像if( argc < 4 ){ printf("** Error. Usage: ./compareHist_Demo <image_settings0> <image_setting1> <image_settings2>\n");return -1;}src_base = imread( argv[1], 1 );src_test1 = imread( argv[2], 1 );src_test2 = imread( argv[3], 1 );/// 转换到 HSVcvtColor( src_base, hsv_base, CV_BGR2HSV );cvtColor( src_test1, hsv_test1, CV_BGR2HSV );cvtColor( src_test2, hsv_test2, CV_BGR2HSV );hsv_half_down = hsv_base( Range( hsv_base.rows/2, hsv_base.rows - 1 ), Range( 0, hsv_base.cols - 1 ) );/// 对hue通道使用30个bin,对saturatoin通道使用32个binint h_bins = 50; int s_bins = 60;int histSize[] = { h_bins, s_bins };// hue的取值范围从0到256, saturation取值范围从0到180float h_ranges[] = { 0, 256 };float s_ranges[] = { 0, 180 };const float* ranges[] = { h_ranges, s_ranges };// 使用第0和第1通道int channels[] = { 0, 1 };/// 直方图MatND hist_base;MatND hist_half_down;MatND hist_test1;MatND hist_test2;/// 计算HSV图像的直方图calcHist( &hsv_base, 1, channels, Mat(), hist_base, 2, histSize, ranges, true, false );normalize( hist_base, hist_base, 0, 1, NORM_MINMAX, -1, Mat() );calcHist( &hsv_half_down, 1, channels, Mat(), hist_half_down, 2, histSize, ranges, true, false );normalize( hist_half_down, hist_half_down, 0, 1, NORM_MINMAX, -1, Mat() );calcHist( &hsv_test1, 1, channels, Mat(), hist_test1, 2, histSize, ranges, true, false );normalize( hist_test1, hist_test1, 0, 1, NORM_MINMAX, -1, Mat() );calcHist( &hsv_test2, 1, channels, Mat(), hist_test2, 2, histSize, ranges, true, false );normalize( hist_test2, hist_test2, 0, 1, NORM_MINMAX, -1, Mat() );///应用不同的直方图对比方法for( int i = 0; i < 4; i++ ){ int compare_method = i;double base_base = compareHist( hist_base, hist_base, compare_method );double base_half = compareHist( hist_base, hist_half_down, compare_method );double base_test1 = compareHist( hist_base, hist_test1, compare_method );double base_test2 = compareHist( hist_base, hist_test2, compare_method );printf( " Method [%d] Perfect, Base-Half, Base-Test(1), Base-Test(2) : %f, %f, %f, %f \n", i, base_base, base_half , base_test1, base_test2 );}printf( "Done \n" );return 0;}

解释

1. 声明储存基准图像和另外两张对比图像的矩阵( RGB 和 HSV )

   Mat src_base, hsv_base;
   Mat src_test1, hsv_test1;
   Mat src_test2, hsv_test2;
   Mat hsv_half_down;
   

2. 装载基准图像(src_base) 和两张测试图像:

   if( argc < 4 ){ printf("** Error. Usage: ./compareHist_Demo <image_settings0> <image_setting1> <image_settings2>\n");return -1;}src_base = imread( argv[1], 1 );
   src_test1 = imread( argv[2], 1 );
   src_test2 = imread( argv[3], 1 );
   

3. 将图像转化到HSV格式:

   cvtColor( src_base, hsv_base, CV_BGR2HSV );
   cvtColor( src_test1, hsv_test1, CV_BGR2HSV );
   cvtColor( src_test2, hsv_test2, CV_BGR2HSV );
   

4. 同时创建包含基准图像下半部的半身图像(HSV格式):

   hsv_half_down = hsv_base( Range( hsv_base.rows/2, hsv_base.rows - 1 ), Range( 0, hsv_base.cols - 1 ) );
   

5. 初始化计算直方图需要的实参(bins, 范围，通道 H 和 S ).

   int h_bins = 50; int s_bins = 32;
   int histSize[] = { h_bins, s_bins };float h_ranges[] = { 0, 256 };
   float s_ranges[] = { 0, 180 };const float* ranges[] = { h_ranges, s_ranges };int channels[] = { 0, 1 };
   

6. 创建储存直方图的 MatND 实例:

   MatND hist_base;
   MatND hist_half_down;
   MatND hist_test1;
   MatND hist_test2;
   

7. 计算基准图像，两张测试图像，半身基准图像的直方图:

   calcHist( &hsv_base, 1, channels, Mat(), hist_base, 2, histSize, ranges, true, false );
   normalize( hist_base, hist_base, 0, 1, NORM_MINMAX, -1, Mat() );calcHist( &hsv_half_down, 1, channels, Mat(), hist_half_down, 2, histSize, ranges, true, false );
   normalize( hist_half_down, hist_half_down, 0, 1, NORM_MINMAX, -1, Mat() );calcHist( &hsv_test1, 1, channels, Mat(), hist_test1, 2, histSize, ranges, true, false );
   normalize( hist_test1, hist_test1, 0, 1, NORM_MINMAX, -1, Mat() );calcHist( &hsv_test2, 1, channels, Mat(), hist_test2, 2, histSize, ranges, true, false );
   normalize( hist_test2, hist_test2, 0, 1, NORM_MINMAX, -1, Mat() );
   

8. 按顺序使用4种对比标准将基准图像(hist_base)的直方图与其余各直方图进行对比:

   for( int i = 0; i < 4; i++ ){ int compare_method = i;double base_base = compareHist( hist_base, hist_base, compare_method );double base_half = compareHist( hist_base, hist_half_down, compare_method );double base_test1 = compareHist( hist_base, hist_test1, compare_method );double base_test2 = compareHist( hist_base, hist_test2, compare_method );printf( " Method [%d] Perfect, Base-Half, Base-Test(1), Base-Test(2) : %f, %f, %f, %f \n", i, base_base, base_half , base_test1, base_test2 );}
   

结果

1. 使用下列输入图像:

   第一张为基准图像，其余两张为测试图像。同时我们会将基准图像与它自身及其半身图像进行对比。

2. 我们应该会预料到当将基准图像直方图及其自身进行对比时会产生完美的匹配， 当与来源于同一样的背景环境的半身图对比时应该会有比较高的相似度， 当与来自不同亮度光照条件的其余两张测试图像对比时匹配度应该不是很好:

3. 下面显示的是结果数值:

对比标准	基准 - 基准	基准 - 半身	基准 - 测试1	基准 - 测试2
Correlation	1.000000	0.930766	0.182073	0.120447
Chi-square	0.000000	4.940466	21.184536	49.273437
Intersection	24.391548	14.959809	3.889029	5.775088
Bhattacharyya	0.000000	0.222609	0.646576	0.801869

对于 Correlation 和 Intersection 标准, 值越大相似度越大。因此可以看到对于采用这两个方法的对比，*基准 - 基准* 的对比结果值是最大的， 而 基准 - 半身 的匹配则是第二好(跟我们预测的一致)。而另外两种对比标准，则是结果越小相似度越大。 我