Grab Cut学习理解之(3)opencv-grab cut

Graph Cut的目标和背景的模型是灰度直方图,Grab Cut取代为RGB三通道的混合高斯模型GMM;建立模型是为了计算一个像素点分别属于目标和背景的概率,介个是为了建图的时候确定Gibbs能量的区域能量项,即图的t-link的权值。

3.1T-Link

单分布高斯背景模型认为,对一个背景图像,特定像素亮度的分布满足高斯分布,即对背景图像B,(x,y)点的亮度满足:IB(x,y) ~ N(u,d),这样我们的背景模型的每个象素属性包括两个参数:平均值u和方差d。对于一幅给定的图像G,如果 Exp(-(IG(x,y)-u(x,y))^2/(2*d^2)) > T,认为(x,y)是背景点,反之是前景点。

高斯模型就是用高斯概率密度函数(正态分布曲线)精确地量化事物,将一个事物分解为若干的基于高斯概率密度函数(正态分布曲线)形成的模型。

图像灰度直方图反映的是图像中某个灰度值出现的频次,也可以认为是图像灰度概率密度的估计。通过将直方图的多峰特性看作是多个高斯分布的叠加,可以解决图像的分割问题。混合高斯模型GMM使用K(基本为3到5个)个高斯模型来表征图像中各个像素点的特征,在新一帧图像获得后更新混合高斯模型, 用当前图像中的每个像素点与混合高斯模型匹配,如果成功则判定该点为背景点, 否则为前景点。高斯模型,主要由方差和均值两个参数决定。

初始时通过k-mean算法分别把属于目标和背景的像素聚类为K类,即GMM中的K个高斯模型,这时候GMM中每个高斯模型就具有了一些像素样本集,这时候它的参数均值和协方差就可以通过他们的RGB值估计得到,而该高斯分量的权值可以通过属于该高斯分量的像素个数与总的像素个数的比值来确定。

最后我们根据建立的高斯模型来计算T-Link的权值,介个是有公式的,然后博客的公式我不会打,不过zouxy09大神有详细的推导过程http://blog.csdn.net/zouxy09/article/details/8532106

3.2N-Link

n-link的权值也就是边界能量项V,和的Graph Cut的计算方案差不多,体现邻域像素m和n之间不连续的惩罚,如果两邻域像素差别很小,那么它属于同一个目标或者同一背景的可能性就很大,如果他们的差别很大,那说明这两个像素很有可能处于目标和背景的边缘部分,则被分割开的可能性比较大,所以当两邻域像素差别越大,能量越小。

3.3GrabCut

这时候根据这两个权值建立图,然后用maxflow算法对其进行分割。

OpenCV依据"GrabCut" - Interactive Foreground Extraction using Iterated Graph Cuts实现了GrabCut算法。

感谢zouxy09大神的源码注解:

/*M///////////////////////////////////////////////////////////////////////////////////////
//
//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
//  By downloading, copying, installing or using the software you agree to this license.
//  If you do not agree to this license, do not download, install,
//  copy or use the software.
//
//
//                        Intel License Agreement
//                For Open Source Computer Vision Library
//
// Copyright (C) 2000, Intel Corporation, all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
//   * Redistribution's of source code must retain the above copyright notice,
//     this list of conditions and the following disclaimer.
//
//   * Redistribution's in binary form must reproduce the above copyright notice,
//     this list of conditions and the following disclaimer in the documentation
//     and/or other materials provided with the distribution.
//
//   * The name of Intel Corporation may not be used to endorse or promote products
//     derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/

#include "precomp.hpp"
#include "gcgraph.hpp"
#include <limits>

using namespace cv;

/*
This is implementation of image segmentation algorithm GrabCut described in
"GrabCut — Interactive Foreground Extraction using Iterated Graph Cuts".
Carsten Rother, Vladimir Kolmogorov, Andrew Blake.
 */

/*
 GMM - Gaussian Mixture Model
*/
class GMM
{
public:
    static const int componentsCount = 5;//5个高斯模型

    GMM( Mat& _model );
    double operator()( const Vec3d color ) const;
    double operator()( int ci, const Vec3d color ) const;
    int whichComponent( const Vec3d color ) const;

    void initLearning();
    void addSample( int ci, const Vec3d color );
    void endLearning();

private:
    void calcInverseCovAndDeterm( int ci );
    Mat model;
    double* coefs;//权值
    double* mean;//均值
    double* cov;//协方差矩阵

    double inverseCovs[componentsCount][3][3]; //协方差的逆矩阵
    double covDeterms[componentsCount];  //协方差的行列式

    double sums[componentsCount][3];
    double prods[componentsCount][3][3];
    int sampleCounts[componentsCount];
    int totalSampleCount;
};

//背景和前景各有一个对应的GMM(混合高斯模型)
GMM::GMM( Mat& _model )
{
	//一个像素的(唯一对应)高斯模型的参数个数或者说一个高斯模型的参数个数
	//一个像素RGB三个通道值,故3个均值,3*3的协方差矩阵,共用一个权值
    const int modelSize = 3/*mean*/ + 9/*covariance*/ + 1/*component weight*/;
    if( _model.empty() )
    {
		//一个GMM共有componentsCount个高斯模型,一个高斯模型有modelSize个模型参数
        _model.create( 1, modelSize*componentsCount, CV_64FC1 );
        _model.setTo(Scalar(0));
    }
    else if( (_model.type() != CV_64FC1) || (_model.rows != 1) || (_model.cols != modelSize*componentsCount) )
        CV_Error( CV_StsBadArg, "_model must have CV_64FC1 type, rows == 1 and cols == 13*componentsCount" );//opencv的错误提示信息

    model = _model;

	//注意这些模型参数的存储方式:先排完componentsCount个coefs,再3*componentsCount个mean。
	//再3*3*componentsCount个cov。
    coefs = model.ptr<double>(0);  //GMM的每个像素的高斯模型的权值变量起始存储指针
    mean = coefs + componentsCount; //均值变量起始存储指针
    cov = mean + 3*componentsCount;  //协方差变量起始存储指针

    for( int ci = 0; ci < componentsCount; ci++ )
        if( coefs[ci] > 0 )
	//计算GMM中第ci个高斯模型的协方差的逆Inverse和行列式Determinant
	//为了后面计算每个像素属于该高斯模型的概率(也就是数据能量项)
        	calcInverseCovAndDeterm( ci );
}

//计算一个像素(由color=(B,G,R)三维double型向量来表示)属于这个GMM混合高斯模型的概率。
//也就是把这个像素像素属于componentsCount个高斯模型的概率与对应的权值相乘再相加,
//具体见论文的公式(10)。结果从res返回。
//这个相当于计算Gibbs能量的第一个能量项(取负后)。
double GMM::operator()( const Vec3d color ) const
{
    double res = 0;
    for( int ci = 0; ci < componentsCount; ci++ )
        res += coefs[ci] * (*this)(ci, color );
    return res;
}

//计算一个像素(由color=(B,G,R)三维double型向量来表示)属于第ci个高斯模型的概率。
//具体过程,即高阶的高斯密度模型计算式,具体见论文的公式(10)。结果从res返回
double GMM::operator()( int ci, const Vec3d color ) const
{
    double res = 0;
    if( coefs[ci] > 0 )
    {
        CV_Assert( covDeterms[ci] > std::numeric_limits<double>::epsilon() );
        Vec3d diff = color;
        double* m = mean + 3*ci;
        diff[0] -= m[0]; diff[1] -= m[1]; diff[2] -= m[2];
        double mult = diff[0]*(diff[0]*inverseCovs[ci][0][0] + diff[1]*inverseCovs[ci][1][0] + diff[2]*inverseCovs[ci][2][0])
                   + diff[1]*(diff[0]*inverseCovs[ci][0][1] + diff[1]*inverseCovs[ci][1][1] + diff[2]*inverseCovs[ci][2][1])
                   + diff[2]*(diff[0]*inverseCovs[ci][0][2] + diff[1]*inverseCovs[ci][1][2] + diff[2]*inverseCovs[ci][2][2]);
        res = 1.0f/sqrt(covDeterms[ci]) * exp(-0.5f*mult);
    }
    return res;
}

//返回这个像素最有可能属于GMM中的哪个高斯模型(概率最大的那个)
int GMM::whichComponent( const Vec3d color ) const
{
    int k = 0;
    double max = 0;

    for( int ci = 0; ci < componentsCount; ci++ )
    {
        double p = (*this)( ci, color );
        if( p > max )
        {
            k = ci;  //找到概率最大的那个,或者说计算结果最大的那个
            max = p;
        }
    }
    return k;
}

//GMM参数学习前的初始化,主要是对要求和的变量置零
void GMM::initLearning()
{
    for( int ci = 0; ci < componentsCount; ci++)
    {
        sums[ci][0] = sums[ci][1] = sums[ci][2] = 0;//rgb三通道嘛
        prods[ci][0][0] = prods[ci][0][1] = prods[ci][0][2] = 0;
        prods[ci][1][0] = prods[ci][1][1] = prods[ci][1][2] = 0;
        prods[ci][2][0] = prods[ci][2][1] = prods[ci][2][2] = 0;
        sampleCounts[ci] = 0;
    }
    totalSampleCount = 0;
}

//增加样本,即为前景或者背景GMM的第ci个高斯模型的像素集(这个像素集是来用估
//计计算这个高斯模型的参数的)增加样本像素。计算加入color这个像素后,像素集
//中所有像素的RGB三个通道的和sums(用来计算均值),还有它的prods(用来计算协方差),
//并且记录这个像素集的像素个数和总的像素个数(用来计算这个高斯模型的权值)。
void GMM::addSample( int ci, const Vec3d color )
{
    sums[ci][0] += color[0]; sums[ci][1] += color[1]; sums[ci][2] += color[2];
    prods[ci][0][0] += color[0]*color[0]; prods[ci][0][1] += color[0]*color[1]; prods[ci][0][2] += color[0]*color[2];
    prods[ci][1][0] += color[1]*color[0]; prods[ci][1][1] += color[1]*color[1]; prods[ci][1][2] += color[1]*color[2];
    prods[ci][2][0] += color[2]*color[0]; prods[ci][2][1] += color[2]*color[1]; prods[ci][2][2] += color[2]*color[2];
    sampleCounts[ci]++;
    totalSampleCount++;
}

//从图像数据中学习GMM的参数:每一个高斯分量的权值、均值和协方差矩阵;
//这里相当于论文中“Iterative minimisation”的step 2
void GMM::endLearning()
{
    const double variance = 0.01;
    for( int ci = 0; ci < componentsCount; ci++ )
    {
        int n = sampleCounts[ci]; //第ci个高斯模型的样本像素个数
        if( n == 0 )
            coefs[ci] = 0;
        else
        {
            //计算第ci个高斯模型的权值系数
	    coefs[ci] = (double)n/totalSampleCount; 

            //计算第ci个高斯模型的均值
	    double* m = mean + 3*ci;
            m[0] = sums[ci][0]/n; m[1] = sums[ci][1]/n; m[2] = sums[ci][2]/n;

            //计算第ci个高斯模型的协方差
	    double* c = cov + 9*ci;
            c[0] = prods[ci][0][0]/n - m[0]*m[0]; c[1] = prods[ci][0][1]/n - m[0]*m[1]; c[2] = prods[ci][0][2]/n - m[0]*m[2];
            c[3] = prods[ci][1][0]/n - m[1]*m[0]; c[4] = prods[ci][1][1]/n - m[1]*m[1]; c[5] = prods[ci][1][2]/n - m[1]*m[2];
            c[6] = prods[ci][2][0]/n - m[2]*m[0]; c[7] = prods[ci][2][1]/n - m[2]*m[1]; c[8] = prods[ci][2][2]/n - m[2]*m[2];

            //计算第ci个高斯模型的协方差的行列式
	    double dtrm = c[0]*(c[4]*c[8]-c[5]*c[7]) - c[1]*(c[3]*c[8]-c[5]*c[6]) + c[2]*(c[3]*c[7]-c[4]*c[6]);
            if( dtrm <= std::numeric_limits<double>::epsilon() )
            {
                //相当于如果行列式小于等于0,(对角线元素)增加白噪声,避免其变
		//为退化(降秩)协方差矩阵(不存在逆矩阵,但后面的计算需要计算逆矩阵)。
		// Adds the white noise to avoid singular covariance matrix.
                c[0] += variance;
                c[4] += variance;
                c[8] += variance;
            }

	    //计算第ci个高斯模型的协方差的逆Inverse和行列式Determinant
            calcInverseCovAndDeterm(ci);
        }
    }
}

//计算协方差的逆Inverse和行列式Determinant
void GMM::calcInverseCovAndDeterm( int ci )
{
    if( coefs[ci] > 0 )
    {
		//取第ci个高斯模型的协方差的起始指针
        double *c = cov + 9*ci;
        double dtrm =
              covDeterms[ci] = c[0]*(c[4]*c[8]-c[5]*c[7]) - c[1]*(c[3]*c[8]-c[5]*c[6])
								+ c[2]*(c[3]*c[7]-c[4]*c[6]);

        //在C++中,每一种内置的数据类型都拥有不同的属性, 使用<limits>库可以获
	//得这些基本数据类型的数值属性。因为浮点算法的截断,所以使得,当a=2,
	//b=3时 10*a/b == 20/b不成立。那怎么办呢?
	//这个小正数(epsilon)常量就来了,小正数通常为可用给定数据类型的
	//大于1的最小值与1之差来表示。若dtrm结果不大于小正数,那么它几乎为零。
	//所以下式保证dtrm>0,即行列式的计算正确(协方差对称正定,故行列式大于0)。
	CV_Assert( dtrm > std::numeric_limits<double>::epsilon() );
	//三阶方阵的求逆
        inverseCovs[ci][0][0] =  (c[4]*c[8] - c[5]*c[7]) / dtrm;
        inverseCovs[ci][1][0] = -(c[3]*c[8] - c[5]*c[6]) / dtrm;
        inverseCovs[ci][2][0] =  (c[3]*c[7] - c[4]*c[6]) / dtrm;
        inverseCovs[ci][0][1] = -(c[1]*c[8] - c[2]*c[7]) / dtrm;
        inverseCovs[ci][1][1] =  (c[0]*c[8] - c[2]*c[6]) / dtrm;
        inverseCovs[ci][2][1] = -(c[0]*c[7] - c[1]*c[6]) / dtrm;
        inverseCovs[ci][0][2] =  (c[1]*c[5] - c[2]*c[4]) / dtrm;
        inverseCovs[ci][1][2] = -(c[0]*c[5] - c[2]*c[3]) / dtrm;
        inverseCovs[ci][2][2] =  (c[0]*c[4] - c[1]*c[3]) / dtrm;
    }
}

//计算beta,也就是Gibbs能量项中的第二项(平滑项)中的指数项的beta,用来调整
//高或者低对比度时,两个邻域像素的差别的影响的,例如在低对比度时,两个邻域
//像素的差别可能就会比较小,这时候需要乘以一个较大的beta来放大这个差别,
//在高对比度时,则需要缩小本身就比较大的差别。
//所以我们需要分析整幅图像的对比度来确定参数beta,具体的见论文公式(5)。
/*
  Calculate beta - parameter of GrabCut algorithm.
  beta = 1/(2*avg(sqr(||color[i] - color[j]||)))
*/
static double calcBeta( const Mat& img )
{
    double beta = 0;
    for( int y = 0; y < img.rows; y++ )
    {
        for( int x = 0; x < img.cols; x++ )
        {
	   //计算四个方向邻域两像素的差别,也就是欧式距离或者说二阶范数
	   //(当所有像素都算完后,就相当于计算八邻域的像素差了)
            Vec3d color = img.at<Vec3b>(y,x);
            if( x>0 ) // left  > 0的判断是为了避免在图像边界的时候还计算,导致越界
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y,x-1);//typedef Vec<uchar, 3> Vec3b;
                beta += diff.dot(diff);  //矩阵的点乘,也就是各个元素平方的和
            }
            if( y>0 && x>0 ) // upleft
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y-1,x-1);
                beta += diff.dot(diff);
            }
            if( y>0 ) // up
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y-1,x);
                beta += diff.dot(diff);
            }
            if( y>0 && x<img.cols-1) // upright
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y-1,x+1);
                beta += diff.dot(diff);
            }
        }
    }
    if( beta <= std::numeric_limits<double>::epsilon() )
        beta = 0;
    else
        beta = 1.f / (2 * beta/(4*img.cols*img.rows - 3*img.cols - 3*img.rows + 2) ); //论文公式(5)

    return beta;
}

/*计算图每个非端点顶点(也就是每个像素作为图的一个顶点,不包括源点s和汇点t)与邻域顶点
  的边的权值。由于是无向图,我们计算的是八邻域,那么对于一个顶点,我们计算四个方向就行,
  在其他的顶点计算的时候,会把剩余那四个方向的权值计算出来。这样整个图算完后,每个顶点
  与八邻域的顶点的边的权值就都计算出来了。
  这个相当于计算Gibbs能量的第二个能量项(平滑项),具体见论文中公式(4)

  Calculate weights of noterminal vertices of graph.
  beta and gamma - parameters of GrabCut algorithm.
 */
static void calcNWeights( const Mat& img, Mat& leftW, Mat& upleftW, Mat& upW,
							Mat& uprightW, double beta, double gamma )
{
    //gammaDivSqrt2相当于公式(4)中的gamma * dis(i,j)^(-1),那么可以知道,
	//当i和j是垂直或者水平关系时,dis(i,j)=1,当是对角关系时,dis(i,j)=sqrt(2.0f)。
	//具体计算时,看下面就明白了
	const double gammaDivSqrt2 = gamma / std::sqrt(2.0f);
	//每个方向的边的权值通过一个和图大小相等的Mat来保存
    leftW.create( img.rows, img.cols, CV_64FC1 );
    upleftW.create( img.rows, img.cols, CV_64FC1 );
    upW.create( img.rows, img.cols, CV_64FC1 );
    uprightW.create( img.rows, img.cols, CV_64FC1 );

    for( int y = 0; y < img.rows; y++ )
    {
        for( int x = 0; x < img.cols; x++ )
        {
            Vec3d color = img.at<Vec3b>(y,x);
            if( x-1>=0 ) // left  //避免图的边界
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y,x-1);
                leftW.at<double>(y,x) = gamma * exp(-beta*diff.dot(diff));
            }
            else
                leftW.at<double>(y,x) = 0;
            if( x-1>=0 && y-1>=0 ) // upleft
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y-1,x-1);
                upleftW.at<double>(y,x) = gammaDivSqrt2 * exp(-beta*diff.dot(diff));
            }
            else
                upleftW.at<double>(y,x) = 0;
            if( y-1>=0 ) // up
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y-1,x);
                upW.at<double>(y,x) = gamma * exp(-beta*diff.dot(diff));
            }
            else
                upW.at<double>(y,x) = 0;
            if( x+1<img.cols && y-1>=0 ) // upright
            {
                Vec3d diff = color - (Vec3d)img.at<Vec3b>(y-1,x+1);
                uprightW.at<double>(y,x) = gammaDivSqrt2 * exp(-beta*diff.dot(diff));
            }
            else
                uprightW.at<double>(y,x) = 0;
        }
    }
}

//检查mask的正确性。mask为通过用户交互或者程序设定的,它是和图像大小一样的单通道灰度图,
//每个像素只能取GC_BGD or GC_FGD or GC_PR_BGD or GC_PR_FGD 四种枚举值,分别表示该像素
//(用户或者程序指定)属于背景、前景、可能为背景或者可能为前景像素。具体的参考:
//ICCV2001“Interactive Graph Cuts for Optimal Boundary & Region Segmentation of Objects in N-D Images”
//Yuri Y. Boykov Marie-Pierre Jolly
/*
  Check size, type and element values of mask matrix.
 */
static void checkMask( const Mat& img, const Mat& mask )
{
    if( mask.empty() )
        CV_Error( CV_StsBadArg, "mask is empty" );
    if( mask.type() != CV_8UC1 )
        CV_Error( CV_StsBadArg, "mask must have CV_8UC1 type" );
    if( mask.cols != img.cols || mask.rows != img.rows )
        CV_Error( CV_StsBadArg, "mask must have as many rows and cols as img" );
    for( int y = 0; y < mask.rows; y++ )
    {
        for( int x = 0; x < mask.cols; x++ )
        {
            uchar val = mask.at<uchar>(y,x);
            if( val!=GC_BGD && val!=GC_FGD && val!=GC_PR_BGD && val!=GC_PR_FGD )
                CV_Error( CV_StsBadArg, "mask element value must be equel"
                    "GC_BGD or GC_FGD or GC_PR_BGD or GC_PR_FGD" );
        }
    }
}

//通过用户框选目标rect来创建mask,rect外的全部作为背景,设置为GC_BGD,
//rect内的设置为 GC_PR_FGD(可能为前景)
/*
  Initialize mask using rectangular.
*/
static void initMaskWithRect( Mat& mask, Size imgSize, Rect rect )
{
    mask.create( imgSize, CV_8UC1 );
    mask.setTo( GC_BGD );

    rect.x = max(0, rect.x);
    rect.y = max(0, rect.y);
    rect.width = min(rect.width, imgSize.width-rect.x);
    rect.height = min(rect.height, imgSize.height-rect.y);

    (mask(rect)).setTo( Scalar(GC_PR_FGD) );
}

//通过k-means算法来初始化背景GMM和前景GMM模型
/*
  Initialize GMM background and foreground models using kmeans algorithm.
*/
static void initGMMs( const Mat& img, const Mat& mask, GMM& bgdGMM, GMM& fgdGMM )
{
    const int kMeansItCount = 10;  //kMeans迭代次数
    const int kMeansType = KMEANS_PP_CENTERS; //Use kmeans++ center initialization by Arthur and Vassilvitskii

    Mat bgdLabels, fgdLabels; //记录背景和前景的像素样本集中每个像素对应GMM的哪个高斯模型,论文中的kn
    vector<Vec3f> bgdSamples, fgdSamples; //背景和前景的像素样本集
    Point p;
    for( p.y = 0; p.y < img.rows; p.y++ )
    {
        for( p.x = 0; p.x < img.cols; p.x++ )
        {
            //mask中标记为GC_BGD和GC_PR_BGD的像素都作为背景的样本像素
		if( mask.at<uchar>(p) == GC_BGD || mask.at<uchar>(p) == GC_PR_BGD )
                bgdSamples.push_back( (Vec3f)img.at<Vec3b>(p) );
            else // GC_FGD | GC_PR_FGD
                fgdSamples.push_back( (Vec3f)img.at<Vec3b>(p) );
        }
    }
    CV_Assert( !bgdSamples.empty() && !fgdSamples.empty() );

//Mat(int rows, int cols, int type, void* data, size_t step=AUTO_STEP)用的是这个重载吧...就是数据不拷贝,只是申请一个头
    Mat _bgdSamples( (int)bgdSamples.size(), 3, CV_32FC1, &bgdSamples[0][0] );
    kmeans( _bgdSamples, GMM::componentsCount, bgdLabels,
            TermCriteria( CV_TERMCRIT_ITER, kMeansItCount, 0.0), 0, kMeansType );
/*
	double kmeans(InputArray samples, int clusterCount, InputOutputArray labels, TermCriteria criteria, int attempts, int flags, OutputArray centers=noArray() )
	_bgdSamples为:一行是一个样本
	componentsCount聚类数
	!! bgdLabels为kmeans的输出,表示每一个样本的类的标签,是一个整数,从0开始的索引
	TermCriteria类为迭代终止条件,该类变量需要3个参数,一个是类型,第二个参数为迭代的最大次数,最后一个是特定的阈值。
	类型有CV_TERMCRIT_ITER、CV_TERMCRIT_EPS、CV_TERMCRIT_ITER+CV_TERMCRIT_EPS,
	分别代表着迭代终止条件为达到最大迭代次数终止,迭代到精度终止,或者两者都作为迭代终止条件。
	attempts-Flag to specify the number of times, the algorithm is executed using different initial labelings.
*/
    Mat _fgdSamples( (int)fgdSamples.size(), 3, CV_32FC1, &fgdSamples[0][0] );
    kmeans( _fgdSamples, GMM::componentsCount, fgdLabels,
            TermCriteria( CV_TERMCRIT_ITER, kMeansItCount, 0.0), 0, kMeansType );

    //经过上面的步骤后,每个像素所属的高斯模型(就是类别)就确定的了,然后就可以估计GMM中每个高斯模型的参数了。
    bgdGMM.initLearning();
    for( int i = 0; i < (int)bgdSamples.size(); i++ )
        bgdGMM.addSample( bgdLabels.at<int>(i,0), bgdSamples[i] );//背景样本按类别加入背景模型辣
    bgdGMM.endLearning();

    fgdGMM.initLearning();
    for( int i = 0; i < (int)fgdSamples.size(); i++ )
        fgdGMM.addSample( fgdLabels.at<int>(i,0), fgdSamples[i] );
    fgdGMM.endLearning();
}

//论文中:迭代最小化算法step 1:为每个像素分配GMM中所属的高斯模型,kn保存在Mat compIdxs中
/*
  Assign GMMs components for each pixel.
*/
static void assignGMMsComponents( const Mat& img, const Mat& mask, const GMM& bgdGMM,
									const GMM& fgdGMM, Mat& compIdxs )
{
    Point p;
    for( p.y = 0; p.y < img.rows; p.y++ )
    {
        for( p.x = 0; p.x < img.cols; p.x++ )
        {
            Vec3d color = img.at<Vec3b>(p);
	   //通过mask来判断该像素属于背景像素还是前景像素,再判断它属于前景或者背景GMM中的哪个高斯分量
            compIdxs.at<int>(p) = mask.at<uchar>(p) == GC_BGD || mask.at<uchar>(p) == GC_PR_BGD ?
                bgdGMM.whichComponent(color) : fgdGMM.whichComponent(color);
        }
    }
}

//论文中:迭代最小化算法step 2:从每个高斯模型的像素样本集中学习每个高斯模型的参数
/*
  Learn GMMs parameters.
*/
static void learnGMMs( const Mat& img, const Mat& mask, const Mat& compIdxs, GMM& bgdGMM, GMM& fgdGMM )
{
    bgdGMM.initLearning();
    fgdGMM.initLearning();
    Point p;
    for( int ci = 0; ci < GMM::componentsCount; ci++ )
    {
        for( p.y = 0; p.y < img.rows; p.y++ )
        {
            for( p.x = 0; p.x < img.cols; p.x++ )
            {
                if( compIdxs.at<int>(p) == ci )
                {
                    if( mask.at<uchar>(p) == GC_BGD || mask.at<uchar>(p) == GC_PR_BGD )
                        bgdGMM.addSample( ci, img.at<Vec3b>(p) );
                    else
                        fgdGMM.addSample( ci, img.at<Vec3b>(p) );
                }
            }
        }
    }
    bgdGMM.endLearning();
    fgdGMM.endLearning();
}

//通过计算得到的能量项构建图,图的顶点为像素点,图的边由两部分构成,
//一类边是:每个顶点与Sink汇点t(代表背景)和源点Source(代表前景)连接的边,
//这类边的权值通过Gibbs能量项的第一项能量项来表示。
//另一类边是:每个顶点与其邻域顶点连接的边,这类边的权值通过Gibbs能量项的第二项能量项来表示。
/*
  Construct GCGraph
*/
static void constructGCGraph( const Mat& img, const Mat& mask, const GMM& bgdGMM, const GMM& fgdGMM, double lambda,
                       const Mat& leftW, const Mat& upleftW, const Mat& upW, const Mat& uprightW,
                       GCGraph<double>& graph )
{
    int vtxCount = img.cols*img.rows;  //顶点数,每一个像素是一个顶点
    int edgeCount = 2*(4*vtxCount - 3*(img.cols + img.rows) + 2);  //边数,需要考虑图边界的边的缺失
    //通过顶点数和边数创建图。这些类型声明和函数定义请参考gcgraph.hpp
	graph.create(vtxCount, edgeCount);
    Point p;
    for( p.y = 0; p.y < img.rows; p.y++ )
    {
        for( p.x = 0; p.x < img.cols; p.x++)
        {
            // add node
            int vtxIdx = graph.addVtx();  //返回这个顶点在图中的索引
            Vec3b color = img.at<Vec3b>(p);

            // set t-weights
            //计算每个顶点与Sink汇点t(代表背景)和源点Source(代表前景)连接的权值。
	   		//也即计算Gibbs能量(每一个像素点作为背景像素或者前景像素)的第一个能量项
	    	double fromSource, toSink;
            if( mask.at<uchar>(p) == GC_PR_BGD || mask.at<uchar>(p) == GC_PR_FGD )
            {

                //对每一个像素计算其作为背景像素或者前景像素的概率,bgdGMM(color) 即color属于bgdGMM的概率计算
                //然后计算齐第一个能量项,作为分别与t和s点的连接权值

				fromSource = -log( bgdGMM(color) );
                toSink = -log( fgdGMM(color) );
            }
            else if( mask.at<uchar>(p) == GC_BGD )
            {
                //对于确定为背景的像素点,它与Source点(前景)的连接为0,与Sink点的连接为lambda
		fromSource = 0;
                toSink = lambda;
            }
            else // GC_FGD
            {
                fromSource = lambda;
                toSink = 0;
            }
	   		//设置该顶点vtxIdx分别与Source点和Sink点的连接权值
            graph.addTermWeights( vtxIdx, fromSource, toSink );

            // set n-weights  n-links
            //计算两个邻域顶点之间连接的权值。
	   		//也即计算Gibbs能量的第二个能量项(平滑项)
	    	if( p.x>0 )
            {
                double w = leftW.at<double>(p);
                graph.addEdges( vtxIdx, vtxIdx-1, w, w );//上一个点肯定是当前点的左边
            }
            if( p.x>0 && p.y>0 )
            {
                double w = upleftW.at<double>(p);
                graph.addEdges( vtxIdx, vtxIdx-img.cols-1, w, w );//当前点与左上点添条边
            }
            if( p.y>0 )
            {
                double w = upW.at<double>(p);
                graph.addEdges( vtxIdx, vtxIdx-img.cols, w, w );//当前点与上点添条边
            }
            if( p.x<img.cols-1 && p.y>0 )
            {
                double w = uprightW.at<double>(p);
                graph.addEdges( vtxIdx, vtxIdx-img.cols+1, w, w );//当前点与右上点添条边
            }
        }
    }
}

//论文中:迭代最小化算法step 3:分割估计:最小割或者最大流算法
/*
  Estimate segmentation using MaxFlow algorithm
*/
static void estimateSegmentation( GCGraph<double>& graph, Mat& mask )
{
    //通过最大流算法确定图的最小割,也即完成图像的分割
    graph.maxFlow();
    Point p;
    for( p.y = 0; p.y < mask.rows; p.y++ )
    {
        for( p.x = 0; p.x < mask.cols; p.x++ )
        {
            //通过图分割的结果来更新mask,即最后的图像分割结果。不会更新用户指定为背景或者前景的像素!
	    	if( mask.at<uchar>(p) == GC_PR_BGD || mask.at<uchar>(p) == GC_PR_FGD )
            {
                if( graph.inSourceSegment( p.y*mask.cols+p.x /*vertex index*/ ) )
                    mask.at<uchar>(p) = GC_PR_FGD;
                else
                    mask.at<uchar>(p) = GC_PR_BGD;
            }
        }
    }
}

//最后的成果:提供给外界使用的伟大的API:grabCut
/*
****参数说明:
	img——待分割的源图像,必须是8位3通道(CV_8UC3)图像,在处理的过程中不会被修改;
	mask——掩码图像,如果使用掩码进行初始化,那么mask保存初始化掩码信息;在执行分割
		的时候,也可以将用户交互所设定的前景与背景保存到mask中,然后再传入grabCut函
		数;在处理结束之后,mask中会保存结果。mask只能取以下四种值:
		GCD_BGD(=0),背景;
		GCD_FGD(=1),前景;
		GCD_PR_BGD(=2),可能的背景;
		GCD_PR_FGD(=3),可能的前景。
		如果没有手工标记GCD_BGD或者GCD_FGD,那么结果只会有GCD_PR_BGD或GCD_PR_FGD;
	rect——用于限定需要进行分割的图像范围,只有该矩形窗口内的图像部分才被处理;
	bgdModel——背景模型,如果为null,函数内部会自动创建一个bgdModel;bgdModel必须是
		单通道浮点型(CV_32FC1)图像,且行数只能为1,列数只能为13x5;
	fgdModel——前景模型,如果为null,函数内部会自动创建一个fgdModel;fgdModel必须是
		单通道浮点型(CV_32FC1)图像,且行数只能为1,列数只能为13x5;
	iterCount——迭代次数,必须大于0;
	mode——用于指示grabCut函数进行什么操作,可选的值有:
		GC_INIT_WITH_RECT(=0),用矩形窗初始化GrabCut;
		GC_INIT_WITH_MASK(=1),用掩码图像初始化GrabCut;
		GC_EVAL(=2),执行分割。
*/
void cv::grabCut( InputArray _img, InputOutputArray _mask, Rect rect,
                  InputOutputArray _bgdModel, InputOutputArray _fgdModel,
                  int iterCount, int mode )
{
    Mat img = _img.getMat();
    Mat& mask = _mask.getMatRef();
    Mat& bgdModel = _bgdModel.getMatRef();
    Mat& fgdModel = _fgdModel.getMatRef();

    if( img.empty() )
        CV_Error( CV_StsBadArg, "image is empty" );
    if( img.type() != CV_8UC3 )
        CV_Error( CV_StsBadArg, "image mush have CV_8UC3 type" );

    GMM bgdGMM( bgdModel ), fgdGMM( fgdModel );//建立模型
    Mat compIdxs( img.size(), CV_32SC1 );//单通道的数据类型为32位有符号整数的

    if( mode == GC_INIT_WITH_RECT || mode == GC_INIT_WITH_MASK )
    {
        if( mode == GC_INIT_WITH_RECT )
            initMaskWithRect( mask, img.size(), rect );//仅用框框初始化
        else // flag == GC_INIT_WITH_MASK
            checkMask( img, mask );
        initGMMs( img, mask, bgdGMM, fgdGMM );
    }

    if( iterCount <= 0)
        return;

    if( mode == GC_EVAL )
        checkMask( img, mask );

    const double gamma = 50;
    const double lambda = 9*gamma;
    const double beta = calcBeta( img );

    Mat leftW, upleftW, upW, uprightW;
    calcNWeights( img, leftW, upleftW, upW, uprightW, beta, gamma );

    for( int i = 0; i < iterCount; i++ )
    {
        GCGraph<double> graph;
        assignGMMsComponents( img, mask, bgdGMM, fgdGMM, compIdxs );
        learnGMMs( img, mask, compIdxs, bgdGMM, fgdGMM );
        constructGCGraph(img, mask, bgdGMM, fgdGMM, lambda, leftW, upleftW, upW, uprightW, graph );
        estimateSegmentation( graph, mask );
    }
}

opencv还提供了grab cut的使用例程,也贴上来吧。。

#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"

#include <iostream>

using namespace std;
using namespace cv;

static void help()
{
    cout << "\nThis program demonstrates GrabCut segmentation -- select an object in a region\n"
            "and then grabcut will attempt to segment it out.\n"
            "Call:\n"
            "./grabcut <image_name>\n"
        "\nSelect a rectangular area around the object you want to segment\n" <<
        "\nHot keys: \n"
        "\tESC - quit the program\n"
        "\tr - restore the original image\n"
        "\tn - next iteration\n"
        "\n"
        "\tleft mouse button - set rectangle\n"
        "\n"
        "\tCTRL+left mouse button - set GC_BGD pixels\n"
        "\tSHIFT+left mouse button - set CG_FGD pixels\n"
        "\n"
        "\tCTRL+right mouse button - set GC_PR_BGD pixels\n"
        "\tSHIFT+right mouse button - set CG_PR_FGD pixels\n" << endl;
}

const Scalar RED = Scalar(0,0,255);//struct Scalar定义可存放1—4个数值的结构,其实里面就是个double数组
const Scalar PINK = Scalar(230,130,255);
const Scalar BLUE = Scalar(255,0,0);
const Scalar LIGHTBLUE = Scalar(255,255,160);
const Scalar GREEN = Scalar(0,255,0);

const int BGD_KEY = CV_EVENT_FLAG_CTRLKEY;//鼠标事件标记(ctrl=8)
const int FGD_KEY = CV_EVENT_FLAG_SHIFTKEY;//(shift=16)

static void getBinMask( const Mat& comMask, Mat& binMask )
{
    if( comMask.empty() || comMask.type()!=CV_8UC1 )
        CV_Error( CV_StsBadArg, "comMask is empty or has incorrect type (not CV_8UC1)" );
    if( binMask.empty() || binMask.rows!=comMask.rows || binMask.cols!=comMask.cols )
        binMask.create( comMask.size(), CV_8UC1 );
    binMask = comMask & 1;//好imba的玩法,是图像中所有的像素值与1做与运算,结果要么是1要么是0?这里有点不懂..
}

class GCApplication
{
public:
    enum{ NOT_SET = 0, IN_PROCESS = 1, SET = 2 };
    static const int radius = 2;//半径
    static const int thickness = -1;//密度

    void reset();//顾名思义,重置
    void setImageAndWinName( const Mat& _image, const string& _winName );
    void showImage() const;
    void mouseClick( int event, int x, int y, int flags, void* param );//回调函数
    int nextIter();//下一次迭代
    int getIterCount() const { return iterCount; }
private:
    void setRectInMask();
    void setLblsInMask( int flags, Point p, bool isPr );

    const string* winName;
    const Mat* image;

    //这些参数好多都是grabCut函数的参数
    Mat mask;
    Mat bgdModel, fgdModel;//前景、背景模型

    uchar rectState, lblsState, prLblsState;//区域设定状态,前景背景点设定状态,可能的前景背景点设定状态
    bool isInitialized;

    Rect rect;//选取前景所在区域或者说处理区域
    vector<Point> fgdPxls, bgdPxls, prFgdPxls, prBgdPxls;//前景点,背景点,可能的前景点,可能的背景点
    int iterCount;
};

void GCApplication::reset()
{
    if( !mask.empty() )
        mask.setTo(Scalar::all(GC_BGD));//掩码图像初始化为GCD_BGD(=0),背景
    bgdPxls.clear(); fgdPxls.clear();
    prBgdPxls.clear();  prFgdPxls.clear();

    isInitialized = false;
    rectState = NOT_SET;
    lblsState = NOT_SET;
    prLblsState = NOT_SET;
    iterCount = 0;
}

void GCApplication::setImageAndWinName( const Mat& _image, const string& _winName  )
{
    if( _image.empty() || _winName.empty() )
        return;
    image = &_image;
    winName = &_winName;
    mask.create( image->size(), CV_8UC1);
    reset();
}

void GCApplication::showImage() const
{
    if( image->empty() || winName->empty() )
        return;

    Mat res;
    Mat binMask;
    if( !isInitialized )
        image->copyTo( res );
    else
    {
        getBinMask( mask, binMask );
        image->copyTo( res, binMask );
    }

    vector<Point>::const_iterator it;
    for( it = bgdPxls.begin(); it != bgdPxls.end(); ++it )
        circle( res, *it, radius, BLUE, thickness );
    for( it = fgdPxls.begin(); it != fgdPxls.end(); ++it )
        circle( res, *it, radius, RED, thickness );
    for( it = prBgdPxls.begin(); it != prBgdPxls.end(); ++it )
        circle( res, *it, radius, LIGHTBLUE, thickness );
    for( it = prFgdPxls.begin(); it != prFgdPxls.end(); ++it )
        circle( res, *it, radius, PINK, thickness );

    if( rectState == IN_PROCESS || rectState == SET )
        rectangle( res, Point( rect.x, rect.y ), Point(rect.x + rect.width, rect.y + rect.height ), GREEN, 2);

    imshow( *winName, res );
}

void GCApplication::setRectInMask()
{
    assert( !mask.empty() );
    mask.setTo( GC_BGD );//整个mask图像都先假定为背景点
    rect.x = max(0, rect.x);
    rect.y = max(0, rect.y);
    rect.width = min(rect.width, image->cols-rect.x);
    rect.height = min(rect.height, image->rows-rect.y);
    (mask(rect)).setTo( Scalar(GC_PR_FGD) );//rect范围都设定为可能的前景点(=3)
}

void GCApplication::setLblsInMask( int flags, Point p, bool isPr )
{
    vector<Point> *bpxls, *fpxls;
    uchar bvalue, fvalue;
    if( !isPr )//确定的前景或背景点
    {
        bpxls = &bgdPxls;
        fpxls = &fgdPxls;
        bvalue = GC_BGD;//0
        fvalue = GC_FGD;//1
    }
    else
    {
        bpxls = &prBgdPxls;
        fpxls = &prFgdPxls;
        bvalue = GC_PR_BGD;//2
        fvalue = GC_PR_FGD;//3
    }
    if( flags & BGD_KEY )
    {
        bpxls->push_back(p);
        circle( mask, p, radius, bvalue, thickness );
    }
    if( flags & FGD_KEY )
    {
        fpxls->push_back(p);
        circle( mask, p, radius, fvalue, thickness );
    }
}

//鼠标函数响应
void GCApplication::mouseClick( int event, int x, int y, int flags, void* )
{
    // TODO add bad args check
    switch( event )//鼠标事件判断
    {
    case CV_EVENT_LBUTTONDOWN: // set rect or GC_BGD(GC_FGD) labels,背景标签设定
        {
            bool isb = (flags & BGD_KEY) != 0,//判断ctrl或者shift是否按下
                 isf = (flags & FGD_KEY) != 0;
            if( rectState == NOT_SET && !isb && !isf )//目标区域未设定
            {
                rectState = IN_PROCESS;
                rect = Rect( x, y, 1, 1 );
            }
            if ( (isb || isf) && rectState == SET )//目标区域设定了并且ctrl或者shift按键按下,则设定背景点或者前景点
                lblsState = IN_PROCESS;
        }
        break;
    case CV_EVENT_RBUTTONDOWN: // set GC_PR_BGD(GC_PR_FGD) labels
        {
            bool isb = (flags & BGD_KEY) != 0,
                 isf = (flags & FGD_KEY) != 0;
            if ( (isb || isf) && rectState == SET )
                prLblsState = IN_PROCESS;
        }
        break;
    case CV_EVENT_LBUTTONUP:
        if( rectState == IN_PROCESS )
        {
            rect = Rect( Point(rect.x, rect.y), Point(x,y) );
            rectState = SET;
            setRectInMask();//rect内部都设定为可能前景点,外部都背景点
            assert( bgdPxls.empty() && fgdPxls.empty() && prBgdPxls.empty() && prFgdPxls.empty() );//断言()内部语句为真
            showImage();
        }
        if( lblsState == IN_PROCESS )
        {
            setLblsInMask(flags, Point(x,y), false);//根据ctrl或shift是否按下在mask中设定前景背景点
            lblsState = SET;
            showImage();
        }
        break;
    case CV_EVENT_RBUTTONUP:
        if( prLblsState == IN_PROCESS )
        {
            setLblsInMask(flags, Point(x,y), true);//根据ctrl或shift是否按下在mask中设定可能的前景背景点
            prLblsState = SET;
            showImage();
        }
        break;
    case CV_EVENT_MOUSEMOVE:
        if( rectState == IN_PROCESS )
        {
            rect = Rect( Point(rect.x, rect.y), Point(x,y) );
            assert( bgdPxls.empty() && fgdPxls.empty() && prBgdPxls.empty() && prFgdPxls.empty() );
            showImage();
        }
        else if( lblsState == IN_PROCESS )
        {
            setLblsInMask(flags, Point(x,y), false);
            showImage();
        }
        else if( prLblsState == IN_PROCESS )
        {
            setLblsInMask(flags, Point(x,y), true);
            showImage();
        }
        break;
    }
}

//Runs the GrabCut algorithm.返回算法运行迭代的次数
int GCApplication::nextIter()
{
/*
贴一段grabCut函数的opencv官网解释
void grabCut(InputArray image, InputOutputArray mask, Rect rect, InputOutputArray bgdModel, InputOutputArray fgdModel, int iterCount, int mode)

Parameters:
image – Input 8-bit 3-channel image.
mask –
Input/output 8-bit single-channel mask. The mask is initialized by the function when mode is set to GC_INIT_WITH_RECT. Its elements may have one of following values:
GC_BGD defines an obvious background pixels.
GC_FGD defines an obvious foreground (object) pixel.
GC_PR_BGD defines a possible background pixel.
GC_PR_BGD defines a possible foreground pixel.
rect – ROI containing a segmented object. The pixels outside of the ROI are marked as “obvious background”. The parameter is only used when mode==GC_INIT_WITH_RECT .
bgdModel – Temporary array for the background model. Do not modify it while you are processing the same image.
fgdModel – Temporary arrays for the foreground model. Do not modify it while you are processing the same image.
iterCount – Number of iterations the algorithm should make before returning the result. Note that the result can be refined with further calls with mode==GC_INIT_WITH_MASK or mode==GC_EVAL .
mode –
Operation mode that could be one of the following:
GC_INIT_WITH_RECT The function initializes the state and the mask using the provided rectangle. After that it runs iterCount iterations of the algorithm.
GC_INIT_WITH_MASK The function initializes the state using the provided mask. Note that GC_INIT_WITH_RECT and GC_INIT_WITH_MASK can be combined. Then, all the pixels outside of the ROI are automatically initialized with GC_BGD .
GC_EVAL The value means that the algorithm should just resume.

*/
    if( isInitialized )
        grabCut( *image, mask, rect, bgdModel, fgdModel, 1 );
    else
    {
        if( rectState != SET )
            return iterCount;

        if( lblsState == SET || prLblsState == SET )
            grabCut( *image, mask, rect, bgdModel, fgdModel, 1, GC_INIT_WITH_MASK );
        else
            grabCut( *image, mask, rect, bgdModel, fgdModel, 1, GC_INIT_WITH_RECT );

        isInitialized = true;//有初次迭代结果,默认么有
    }
    iterCount++;

    bgdPxls.clear(); fgdPxls.clear();
    prBgdPxls.clear(); prFgdPxls.clear();

    return iterCount;
}

GCApplication gcapp;

static void on_mouse( int event, int x, int y, int flags, void* param )
{
    gcapp.mouseClick( event, x, y, flags, param );
}

int main( int argc, char** argv )
{
    if( argc!=2 )
    {
        help();
        return 1;
    }
    string filename = argv[1];
    if( filename.empty() )
    {
        cout << "\nDurn, couldn't read in " << argv[1] << endl;
        return 1;
    }
    Mat image = imread( filename, 1 );
    if( image.empty() )
    {
        cout << "\n Durn, couldn't read image filename " << filename << endl;
        return 1;
    }

    help();

    const string winName = "image";
    namedWindow( winName, WINDOW_AUTOSIZE );
    setMouseCallback( winName, on_mouse, 0 );//Sets mouse handler for the specified window
    					     //void cvSetMouseCallback(const char* name, CvMouseCallback onMouse, void* param=NULL )
    					     //在winName窗口调用on_mouse,介个on_mouse函数是要有一定格式滴

    gcapp.setImageAndWinName( image, winName );
    gcapp.showImage();

    for(;;)
    {
        int c = waitKey(0);
        switch( (char) c )
        {
        case '\x1b':
            cout << "Exiting ..." << endl;
            goto exit_main;
        case 'r':
            cout << endl;
            gcapp.reset();
            gcapp.showImage();
            break;
        case 'n':
            int iterCount = gcapp.getIterCount();
            cout << "<" << iterCount << "... ";
            int newIterCount = gcapp.nextIter();
            if( newIterCount > iterCount )
            {
                gcapp.showImage();
                cout << iterCount << ">" << endl;
            }
            else
                cout << "rect must be determined>" << endl;
            break;
        }
    }

exit_main:
    destroyWindow( winName );
    return 0;
}
时间: 2024-08-13 03:41:12

Grab Cut学习理解之(3)opencv-grab cut的相关文章

Grab Cut学习理解之(1)new min-cut /max-flow algorithm

参(chao)考(xi)的资料还是很多的,好好学习,天天向上嘛- 1.swsamleo大神:http://blog.csdn.net/swsamleo/article/details/7915316 2.wstcegg大神:http://blog.csdn.net/wstcegg/article/details/39495535 3.zouxy09大神图像分割系列:http://blog.csdn.net/zouxy09/article/details/8532106 4.GraphCuts o

学习理解shell的好办法--编写自己的shell 之二

shell脚本的最简单形式就是一串命令的罗列,shell充当解释器,一条条挨个执行,直到最后一个或遇到退出命令.但这只能做很简单的事情,只是省区了每次都要敲一边命令的时间,要想完成更负责的功能,还要加上这些东西: 1.控制 前面的条件满足了,然后干什么:不满足,干什么. 2.变量 c=a+b, 用一种形式代表另一种形式,就是变量.因为形式不同了,就能用一种不变的表示另一种变化的.比如"编程语言"就可以当一个变量,可以赋值为"C语言","Perl语言&quo

Android中的context的学习理解

Android中Context的学习理解Context是一个抽象基类,通过它getResuources.getAssets and start 其他组件(Activity,Service,broadCast,getSystemService),可以这样理解:Context提供了一个运行环境for App, then app 可以访问资源,才能完成与其他组件,服务的交互,Context定义了一套基本的功能接口or一套规范 //todo

学习理解 makefile

学习理解 makefile 模拟个应用的例子: 有个工程包括头文件 001.h.002.h.003.h.004.h.005.h.006.h.007.h 共7个:程序文件 001.c.002.c.003.c.004.c.005.c.006.c.007.c.008.c.009.c.010.c 共10个文件.看着头大吧,先不关心具体内容. 现在来编译该工程.如下: # cd example/ # gcc 001.c 002.c 003.c 004.c 005.c 006.c 007.c 008.c 0

学习理解shell的好办法--编写自己的shell 之一

本文参考自<Unix/Linux编程实践教程>, 这是一本讲解unix系统编程的书,注重实践,理解难度不大,推荐大家阅读,敲完本书后,对于理解unix系统如何运作会有更深的视角,回过头再学习别的 Linux相关的东西时,感受非常不一样,这是一本可以提高"内功"的书.自己加了些很菜的解释,以便其他小白理解,大牛直接飘过吧,错误之处希望指正. shell是一个管理进程和运行程序的程序,用来人和机器交互 常用的shell如sh,bash,zsh,csh,ksh等都有三个主要功能:

有关meanshift跟踪的理解(在opencv中实现)(转载)

meanshift算法思想其实很简单:利用概率密度的梯度爬升来寻找局部最优.它要做的就是输入一个在图像的范围,然后一直迭代(朝着重心迭代)直到满足你的要求为止.但是他是怎么用于做图像跟踪的呢?这是我自从学习meanshift以来,一直的困惑.而且网上也没有合理的解释.经过这几天的思考,和对反向投影的理解使得我对它的原理有了大致的认识. 在opencv中,进行meanshift其实很简单,输入一张图像(imgProb),再输入一个开始迭代的方框(windowIn)和一个迭代条件(criteria)

深入学习理解java-ThreadLocal

导读 首先,ThreadLocal 不是用来解决共享对象的多线程访问问题的,一般情况下,通过ThreadLocal.set() 到线程中的对象是该线程自己使用的对象,其他线程是不需要访问的,也访问不到的.各个线程中访问的是不同的对象. 另外,说ThreadLocal使得各线程能够保持各自独立的一个对象,并不是通过ThreadLocal.set()来实现的,而是通过每个线程中的new 对象 的操作来创建的对象,每个线程创建一个,不是什么对象的拷贝或副本.通过ThreadLocal.set()将这个

思维探索者:理解了才是真正的学习 理解会带来巨大的好处

http://www.nowamagic.net/librarys/veda/detail/1718先说三件事情: 一个问题其实你可以一直放在脑子里面,利用暗时间对其软泡硬磨,时间足够久你总会有一点新的感悟,问题其实就像那句老话说的那样,不怕贼偷就怕贼惦记,聚精会神的思考一天,也许比不上惦记一个星期(据说数学家庞加莱就特别会惦记问题). 事实上,当你感觉懂了的时候,你至少得反问自己一句,真的懂了吗?当你确信自己真的懂了的时候,你至少得讲给别人听,别人听懂了吗?考察你自己是否真懂了的一个很好的依据

Source Xref 与 JavaDocs 学习理解

最近学习Mybatis的官方文档,看到了[项目文档]一节有很多内容没有见过,做个笔记,理解一下. 没找到java相关代码的解释,其实用下面这个php版本解释,也非常不错. What is SOURCEXref? SOURCEXref is a developer tool that’s designed to ease the process of working on large PHP projects by making it very fast and easy to browse th