- #include <opencv2/core/core.hpp>
- #include <opencv2/highgui/highgui.hpp>
- #include <opencv2/imgproc/imgproc.hpp>
- #include <iostream>
- #pragma comment(lib, "opencv_core231d.lib")
- #pragma comment(lib, "opencv_highgui231d.lib")
- #pragma comment(lib, "opencv_imgproc231d.lib")
- using namespace cv;
- using namespace std;
- #define HIST_DIM1
- int main( int argc, char** argv )
- {
- #ifdef HIST_DIM1
- //----------------------example 1-------------------------------//
- Mat src, dst;
- /// Load image
- src = imread("d:/picture/lena.jpg");
- if( !src.data )
- {
- cout<<"load image failed"<<endl;
- return -1;
- }
- /// Separate the image in 3 places ( R, G and B )
- vector<Mat> rgb_planes;
- #define SHOW_HSV
- #ifdef SHOW_HSV
- Mat hsv;
- cvtColor(src, hsv, COLOR_BGR2HSV);
- split(hsv, rgb_planes );
- #else
- split(src, rgb_planes );
- #endif
- /// Establish the number of bins
- int histSize = 256;
- /// Set the ranges ( for R,G,B) )
- float range[] = { 0, 255 } ;
- const float* histRange = { range };
- bool uniform = true; bool accumulate = false;
- Mat r_hist, g_hist, b_hist;
- /// Compute the histograms:
- calcHist( &rgb_planes[2], 1, 0, Mat(), r_hist, 1, &histSize, &histRange, uniform, accumulate );
- calcHist( &rgb_planes[1], 1, 0, Mat(), g_hist, 1, &histSize, &histRange, uniform, accumulate );
- calcHist( &rgb_planes[0], 1, 0, Mat(), b_hist, 1, &histSize, &histRange, uniform, accumulate );
- // Draw the histograms for R, G and B
- int hist_w = 600; int hist_h = 400;
- int bin_w = cvRound( (double) hist_w/histSize );
- Mat rgb_hist[3];
- for(int i=0; i<3; ++i)
- {
- rgb_hist[i] = Mat(hist_h, hist_w, CV_8UC3, Scalar::all(0));
- }
- Mat histImage(hist_h, hist_w, CV_8UC3, Scalar(0,0,0));
- /// Normalize the result to [ 0, histImage.rows-10]
- normalize(r_hist, r_hist, 0, histImage.rows-10, NORM_MINMAX);
- normalize(g_hist, g_hist, 0, histImage.rows-10, NORM_MINMAX);
- normalize(b_hist, b_hist, 0, histImage.rows-10, NORM_MINMAX);
- /// Draw for each channel
- for( int i = 1; i < histSize; i++ )
- {
- line( histImage, Point( bin_w*(i-1), hist_h-cvRound(r_hist.at<float>(i-1)) ) ,
- Point( bin_w*(i), hist_h-cvRound(r_hist.at<float>(i)) ),
- Scalar( 0, 0, 255), 1);
- line( histImage, Point( bin_w*(i-1), hist_h - cvRound(g_hist.at<float>(i-1)) ) ,
- Point( bin_w*(i), hist_h-cvRound(g_hist.at<float>(i)) ),
- Scalar( 0, 255, 0), 1);
- line( histImage, Point( bin_w*(i-1), hist_h - cvRound(b_hist.at<float>(i-1)) ) ,
- Point( bin_w*(i), hist_h-cvRound(b_hist.at<float>(i)) ),
- Scalar( 255, 0, 0), 1);
- }
- for (int j=0; j<histSize; ++j)
- {
- int val = saturate_cast<int>(r_hist.at<float>(j));
- rectangle(rgb_hist[0], Point(j*2+10, rgb_hist[0].rows), Point((j+1)*2+10, rgb_hist[0].rows-val), Scalar(0,0,255),1,8);
- val = saturate_cast<int>(g_hist.at<float>(j));
- rectangle(rgb_hist[1], Point(j*2+10, rgb_hist[1].rows), Point((j+1)*2+10, rgb_hist[1].rows-val), Scalar(0,255,0),1,8);
- val = saturate_cast<int>(b_hist.at<float>(j));
- rectangle(rgb_hist[2], Point(j*2+10, rgb_hist[2].rows), Point((j+1)*2+10, rgb_hist[2].rows-val), Scalar(255,0,0),1,8);
- }
- /// Display
- namedWindow("calcHist Demo", CV_WINDOW_AUTOSIZE );
- namedWindow("wnd");
- imshow("calcHist Demo", histImage );
- imshow("wnd", src);
- imshow("R", rgb_hist[0]);
- imshow("G", rgb_hist[1]);
- imshow("B", rgb_hist[2]);
- #else
- //----------------------example 2-------------------------------//
- Mat src, hsv;
- if(!(src=imread("d:/picture/lena.bmp")).data)
- return -1;
- cvtColor(src, hsv, CV_BGR2HSV);
- // Quantize the hue to 30 levels
- // and the saturation to 32 levels
- int hbins = 60, sbins = 64;
- int histSize[] = {hbins, sbins};
- // hue varies from 0 to 179, see cvtColor
- float hranges[] = { 0, 180 };
- // saturation varies from 0 (black-gray-white) to
- // 255 (pure spectrum color)
- float sranges[] = { 0, 256};
- const float*ranges[] = { hranges, sranges };
- MatND hist;
- // we compute the histogram from the 0-th and 1-st channels
- int channels[] = {0, 1};
- calcHist( &hsv, 1, channels, Mat(),hist, 2, histSize, ranges,true, false );
- double maxVal=0;
- minMaxLoc(hist, 0, &maxVal, 0, 0);
- int scale = 8;
- Mat histImg = Mat::zeros(sbins*scale, hbins*scale, CV_8UC3);
- for( int h = 0; h < hbins; h++ )
- {
- for( int s = 0; s < sbins; s++ )
- {
- float binVal = hist.at<float>(h, s);
- int intensity = cvRound(binVal*255/maxVal);
- rectangle( histImg, Point(h*scale, s*scale),Point((h+1)*scale-1, (s+1)*scale-1), Scalar::all(intensity), CV_FILLED);
- }
- }
- namedWindow( "Source", 1 );
- imshow( "Source", src );
- namedWindow( "H-S Histogram", 1 );
- imshow( "H-S Histogram", histImg );
- #endif
- //-------------------------------------------------------------------------//
- waitKey(0);
- destroyAllWindows();
- return 0;
- }
上面的例子是对opencv_tutorials以及手册中的计算直方图的程序的修改
其中修改的:
1、原先的程序中对加载的彩色rgb图像的通道有问题(看例子给的图应该是在linux下的,不知道是不是因为linux和windows下加载的不同),在windows下默认加载的通道排列顺序是B-G-R,
原先的程序中是按照R-G-B顺序计算的直方图所以需要变换一下顺序;
2、原先程序的histImage将参数顺序弄错了,该构造函数的第一个参数是rows行数,对应图像的高度,即hist_h,而不是hist_w,这里同时还将大小变换了一下
看着更舒服一些;
下面是对calcHist函数的参数介绍。
calcHist--计算矩阵的直方图函数;
--------------------------------------------------------------------------------------------------------------
- ###---given in manual---###
- void calcHist(const Mat*arrays, int narrays, const int* channels, InputArray mask, OutputArray hist, int dims,
- const int* histSize, const float** ranges, bool uniform=true, boolaccumulate=false)
- void calcHist(const Mat*arrays, int narrays, const int* channels, InputArray mask, SparseMat& hist, int dims,
- const int* histSize, const float** ranges, bool uniform=true, boolaccumulate=false)
- ###---declaration in imgproc.hpp---###
- //! computes the joint dense histogram for a set of images.
- CV_EXPORTS void calcHist( const Mat* images, int nimages, const int* channels, InputArray mask, OutputArray hist, int dims,
- const int* histSize, const float** ranges, bool uniform=true, bool accumulate=false );
- //! computes the joint sparse histogram for a set of images.
- CV_EXPORTS void calcHist( const Mat* images, int nimages, const int* channels, InputArray mask, SparseMat& hist, int dims,
- const int* histSize, const float** ranges, bool uniform=true, bool accumulate=false );
- CV_EXPORTS_W void calcHist( InputArrayOfArrays images, const vector<int>& channels, InputArray mask, OutputArray hist,
- const vector<int>& histSize, const vector<float>& ranges,bool accumulate=false );
--------------------------------------------------------------------------------------------------------------
手册中和头文件中的函数声明参数稍有不同,主要是前两个参数,手册中是array和narrays而头文件声明中是images和nimages,其实是一样,以手册为准:
这里有一个对opencv_tutorials.pdf即opencv教程的一个翻译。
http://www.opencv.org.cn/opencvdoc/2.3.2/html/doc/tutorials/imgproc/histograms/histogram_calculation/histogram_calculation.html
arrays – Source arrays. They all should have the same depth, CV_8U or CV_32F , and the same size. Each of them can have an arbitrary number of channels.
- 源输入(图像)数组,必须是相同深度的CV_8U或者CV_32F(即uchar或者float),相同大小,每一个可以是任意通道的;
[上面的例子1中每次计算一个单通道图像,所以直接对图像取地址赋给了该参数]
narrays – Number of source arrays.
- 源输入数组中的元素个数;
[例子1中只计算一幅图像的直方图,所以这个参数都是1]
channels – List of the dims channels used to compute the histogram. The first array channels are enumerated from 0 to arrays[0].channels()-1 , the second array channels are counted from arrays[0].channels() to arrays[0].channels() + arrays[1].channels()-1, and so on.
- 用来计算直方图的通道维数数组,第一个数组的通道由0到arrays[0].channels()-1列出,第二个数组的通道从arrays[0].channels()到arrays[0].channels()+arrays[1].channels()-1以此类推;
[例子1中为0,即第0个通道??]
mask – Optional mask. If the matrix is not empty, it must be an 8-bit array of the same size as arrays[i]. The non-zero mask elements mark the array elements counted in the histogram.
-可选的掩膜,如果该矩阵不是空的,则必须是8位的并且与arrays[i]的大小相等,掩膜的非零值标记需要在直方图中统计的数组元素;
[例子1中为空的Mat()]
hist – Output histogram, which is a dense or sparse dims -dimensional array.
-输出直方图,是一个稠密或者稀疏的dims维的数组;
[例子1中为保存直方图的Mat]
dims – Histogram dimensionality that must be positive and not greater than CV_MAX_DIMS (equal to 32 in the current OpenCV version).
-直方图的维数,必须为正,并且不大于CV_MAX_DIMS(当前的OpenCV版本中为32,即最大可以统计32维的直方图);
[例子1中为1,因为统计的是每幅单通道图像的灰度直方图]
histSize – Array of histogram sizes in each dimension.
- 用于指出直方图数组每一维的大小的数组,即指出每一维的bin的个数的数组;
[因为例子1只有1维,所以例子1中直接对int取地址作为参数,即该维的bin的个数为256]
ranges – Array of the dims arrays of the histogram bin boundaries in each dimension. When the histogram is uniform ( uniform =true), then for each dimension i it is enough to specify the lower (inclusive) boundary of the 0-th histogram bin and the upper(exclusive) boundary for the last histogram bin histSize[i]-1. That is, in case of a uniform histogram each of ranges[i] is an array of 2 elements. When the histogram is not uniform ( uniform=false ), then each of ranges[i] contains histSize[i]+1 elements:. The array elements, that are not between and , are not counted in the histogram.
- 用于指出直方图每一维的每个bin的上下界范围数组的数组,当直方图是均匀的(uniform =true)时,对每一维i指定直方图的第0个bin的下界(包含即[)L0和最后一个即第histSize[i]-1个bin的上界(不包含的即))U_histSize[i]-1,也就是说对均匀直方图来说,每一个ranges[i]都是一个两个元素的数组【指出该维的上下界】。当直方图不是均匀的时,每一个ranges[i]数组都包含histSize[i]+1个元素:L0,U0=L1,U1=L1,...,U_histSize[i]-2 = L_histSize[i]-1,U_histSize[i]-1.不在L0到U_histSize[i]-1之间的数组元素将不会统计进直方图中;
[在例子1中采用的是均匀直方图,所以范围为0-255]
uniform – Flag indicates that whether the histogram is uniform or not (see above).
- 直方图是否均匀的标志;【指定直方图每个bin统计的是否是相同数量的灰度级】
[例子1中为true]
accumulate – Accumulation flag. If it is set, the histogram is not cleared in the beginning when it is allocated.
This feature enables you to compute a single histogram from several sets of arrays, or to update the histogram in time.
-累加标志;
[单幅图像不进行累计所以例子1中为false]
参数中最难理解的应该就是channels和ranges这两个参数,以及histSize和ranges这两个参数的关系,关于histSize和ranges的关系也就涉及了ranges的意义,关于它们的关系在《学习OpenCV中文版》09.10第一版的page:219-220有比较清楚的说明。
【channels参数,自己也不是很明确,等看看该函数的源码之后再说】。
使用上面第一个例子获得的lena的hsv直方图如下:直接在RGB基础上修改的,所以窗口名字对应R-V,G-S,B-H。
手册中该函数的介绍之后有个例子,是计算图像的2维H-S直方图的,就是上面的例子2(稍微进行了一点修改);
这个例子中的参数分别为:
参数1:&hsv,一幅HSV三通道的彩色图像指针;
参数2:1,因为参数1是一幅图像;
参数3:channels,数组包含两个元素:0,1;--指明要统计的是通道0和通道1的数据??--不确定是否是这样的!
参数4:Mat(),为空,不使用掩膜;
参数5:hist,输出2D直方图,MatND,也就是Mat;
参数6:2,2维直方图;
参数7:histSize,两个元素的数组,指明每一维的bin的个数,上面的例子2中,h的为60,s的为64;
参数8:ranges,指出bin的范围的数组的数组,因为后面的uniform标志为true,也就是均匀直方图,所以每一维由一个两个元素的数组指出上下限;
参数9:true,也就是采用均匀直方图;
参数10:false,不使用累积;
第二个例子给出的2维直方图中,水平的是h分量,垂直的是s分量,下面是lena的H-S直方图图像:
可以看出,h分量也就是H-S直方图的垂直投影集中在0和60左右,对应到hsv空间也就是色相为红色部分,从例子1的h分量直方图以及直观的看lena原始图像也可以看出来,
而s分量,也就是水平投影,集中在中间部分,从例子1的s分量的直方图中也可以看出;
而且如果将这个H-S二维直方图的每一维的bin数量设置的与上面的例1中一样,然后在分别向垂直方向和水平方向投影,获得的两个投影直方图应该与例1中的对应
直方图是一样的。
上次对calcHist的参数进行了分析,并且给出了几个例子,但是对channels参数没搞清楚,今天又写了个例子分析了一下,终于弄明白了。
calcHist函数的channels参数和narrays以及dims共同来确定用于计算直方图的图像;
首先dims是最终的直方图维数,narrays指出了arrays数组中图像的个数,其中每一幅图像都可以是任意通道的【只要最终dims不超过32即可】
如果channels参数为0,则narrays和dims必须相等,否则弹出assert,此时计算直方图的时候取数组中每幅图像的第0通道。
当channels不是0的时候,用于计算直方图的图像是arrays中由channels指定的通道的图像,channels与arrays中的图像的对应关系,如channels的参数说明的,将arrays中的图像从第0幅开始按照通道摊开排列起来,然后channels中的指定的用于计算直方图的就是这些摊开的通道;
假设有arrays中只有一幅三通道的图像image,那么narrays应该为1,如果是想计算3维直方图【最大也只能是3维的】,想将image的通道2作为第一维,通道0作为第二维,通道1作为第三维,则可以将channels设置为channesl={2,0,1};这样calcHist函数计算时就按照这个顺序来统计直方图。
可以看出channels不为0时narrays可以和dims不相等,只要保证arrays中至少有channels指定的通道就可以。
下面是例子和一个运行的图像,还是使用lena图像;
- Mat src, hsv;
- if(!(src=imread("d:/picture/lena.bmp")).data)
- return -1;
- cvtColor(src, hsv, CV_BGR2HSV);
- vector<Mat> hsv_plane;
- split(hsv, hsv_plane);
- Mat inputs[]={hsv_plane[1], hsv_plane[2], hsv_plane[0]};
- vector<Mat> mixmat_plane;
- mixmat_plane.push_back(hsv_plane[2]);
- mixmat_plane.push_back(hsv_plane[0]);
- Mat mixmat;
- merge(mixmat_plane, mixmat);
- Mat mixed[]={mixmat,hsv_plane[1]};
- int vbins = 128, sbins = 128, hbins = 128;
- int histSize[] = {sbins, vbins, hbins};
- float sranges[] = { 0, 256};
- float vranges[] = { 0, 256};
- float hranges[] = { 0, 256};
- const float*ranges[] = {sranges, vranges, hranges};
- MatND hist;
- //#define SINGLE_MAT
- #define MIX_MAT
- #ifdef SINGLE_MAT
- /*
- use one multi-channel mat, channels param gives the channels used;
- 使用多通道的图像计算多维直方图,可以计算1,2,3维的;
- */ int channels[] = {1, 2};
- calcHist(&hsv, 1, channels, Mat(),hist, 2, histSize, ranges,true, false );
- #elif defined MIX_MAT
- /*
- use mix mat array, the first elem is a single channel mat, second is a two channel mat;
- 使用混合通道图像数组,第1个图像是2通道的,第2个是单通道的;
- channels指定每一维对应的通道;
- */
- int channels[] = {1, 2, 0};
- // #define DIM_2
- #ifdef DIM_2
- //统计二维直方图;
- calcHist(mixed, 2, channels, Mat(),hist, 2, histSize, ranges,true, false);
- #else
- //统计三维直方图;
- calcHist(mixed, 2, channels, Mat(),hist, 3, histSize, ranges,true, false);
- #endif
- #else
- /*
- use multi-mat arrays, channels param gives the array mat and its channels used;
- 使用都是单通道图像数组计算2维直方图--也可以计算3维的;
- */
- int channels[] = {2, 1};
- hbins = 1;
- calcHist(inputs, 3, channels, Mat(),hist, 2, histSize, ranges,true, false );
- #endif
- #ifndef MIX_MAT
- double maxVal=0;
- minMaxLoc(hist, 0, 0, 0, 0);//only can process mat that dims<=2--minMaxLoc只能处理2维以下的;
- #endif
- int scale = 4;
- Mat histImg = Mat::zeros(vbins*scale, sbins*scale, CV_8UC3);
- float *hist_sta = new float[sbins];
- float *hist_val = new float[vbins];
- float *hist_hue = new float[hbins];
- memset(hist_val, 0, vbins*sizeof(float));
- memset(hist_sta, 0, sbins*sizeof(float));
- memset(hist_hue, 0, hbins*sizeof(float));
- for( int s = 0; s < sbins; s++ )
- {
- for( int v = 0; v < vbins; v++ )
- {
- for(int h=0; h<hbins; h++)
- {
- #ifdef MIX_MAT
- //-----------------------------------------------------------//
- #ifdef DIM_2
- float binVal = hist.at<float>(s, v);
- #else
- float binVal = hist.at<float>(s, v, h);
- hist_hue[h] += binVal;
- #endif
- //-----------------------------------------------------------//
- #else
- float binVal = hist.at<float>(s, v);
- int intensity = cvRound(binVal*255/maxVal);
- rectangle( histImg, Point(s*scale, v*scale),Point((s+1)*scale-1, (v+1)*scale-1), Scalar::all(intensity), CV_FILLED);
- #endif
- hist_val[v] += binVal;
- hist_sta[s] += binVal;
- }
- }
- }
- //find max bin value;
- double max_sta=.0, max_val=.0,max_hue=.0;
- for(int i=0; i<sbins; ++i)
- {
- if(hist_sta[i]>max_sta)
- max_sta = hist_sta[i];
- }
- for(int i=0; i<vbins; ++i)
- {
- if(hist_val[i]>max_val)
- max_val = hist_val[i];
- }
- for(int i=0; i<hbins; ++i)
- {
- if(hist_hue[i]>max_hue)
- max_hue = hist_hue[i];
- }
- Mat sta_img = Mat::zeros(310, sbins*scale+20, CV_8UC3);
- Mat val_img = Mat::zeros(310, vbins*scale+20, CV_8UC3);
- Mat hue_img = Mat::zeros(310, hbins*scale+20, CV_8UC3);
- for(int i=0; i<sbins; ++i)
- {
- int intensity = cvRound(hist_sta[i]*(sta_img.rows-10)/max_sta);
- rectangle(sta_img, Point(i*scale+10, sta_img.rows-intensity),Point((i+1)*scale-1+10, sta_img.rows-1), Scalar(0,255,0), 1);
- }
- for(int i=0; i<vbins; ++i)
- {
- int intensity = cvRound(hist_val[i]*(val_img.rows-10)/max_val);
- rectangle(val_img, Point(i*scale+10, val_img.rows-intensity),Point((i+1)*scale-1+10, val_img.rows-1), Scalar(0,0,255), 1);
- }
- for(int i=0; i<hbins; ++i)
- {
- int intensity = cvRound(hist_hue[i]*(hue_img.rows-10)/max_hue);
- rectangle(hue_img, Point(i*scale+10, hue_img.rows-intensity),Point((i+1)*scale-1+10, hue_img.rows-1), Scalar(255,0,0), 1);
- }
- namedWindow( "Source");
- imshow( "Source", src );
- namedWindow( "Histogram");
- imshow( "Histogram", histImg );
- namedWindow("dim1");
- imshow("dim1", sta_img);
- namedWindow("dim2");
- imshow("dim2", val_img);
- namedWindow("dim3");
- imshow("dim3", hue_img);
程序中使用了一些宏来控制不同的情况,比较简单一看就明白,毋庸多说。上图中的channels顺序是1,2,0,而图像数组是采用将VH组成一个两通道图像以及S图像放到一个数组中,即混合通道的数组,计算的是3通道的直方图,然后将每一维拆开了,分别显示在dim1-3中。
颜色直方图距离匹配: 方法一,三个通道分别计算直方图然后归一化合成一维向量,计算距离 方法二,三通道直方图合成一维向量,然后归一化,最后计算距离, 方法二的效果要明显好于方法一,略逊于直接计算二维直方图。