常用图像二值化方法

原文地址：http://www.cnblogs.com/skyseraph/archive/2010/12/21/1913058.html

代码/*===============================图像分割=====================================*//*---------------------------------------------------------------------------*//*手动设置阀值*/    IplImage* binaryImg = cvCreateImage(cvSize(w, h),IPL_DEPTH_8U, 1);    cvThreshold(smoothImgGauss,binaryImg,71,255,CV_THRESH_BINARY);     cvNamedWindow("cvThreshold", CV_WINDOW_AUTOSIZE );    cvShowImage( "cvThreshold", binaryImg );//cvReleaseImage(&binaryImg);     /*---------------------------------------------------------------------------*//*自适应阀值  //计算像域邻域的平均灰度，来决定二值化的值*/    IplImage* adThresImg = cvCreateImage(cvSize(w, h),IPL_DEPTH_8U, 1);double max_value=255;int adpative_method=CV_ADAPTIVE_THRESH_GAUSSIAN_C;//CV_ADAPTIVE_THRESH_MEAN_C int threshold_type=CV_THRESH_BINARY;int block_size=3;//阈值的象素邻域大小 int offset=5;//窗口尺寸     cvAdaptiveThreshold(smoothImgGauss,adThresImg,max_value,adpative_method,threshold_type,block_size,offset);    cvNamedWindow("cvAdaptiveThreshold", CV_WINDOW_AUTOSIZE );    cvShowImage( "cvAdaptiveThreshold", adThresImg );    cvReleaseImage(&adThresImg);/*---------------------------------------------------------------------------*//*最大熵阀值分割法*/        IplImage* imgMaxEntropy = cvCreateImage(cvGetSize(imgGrey),IPL_DEPTH_8U,1);    MaxEntropy(smoothImgGauss,imgMaxEntropy);    cvNamedWindow("MaxEntroyThreshold", CV_WINDOW_AUTOSIZE );    cvShowImage( "MaxEntroyThreshold", imgMaxEntropy );//显示图像     cvReleaseImage(&imgMaxEntropy ); /*---------------------------------------------------------------------------*//*基本全局阀值法*/    IplImage* imgBasicGlobalThreshold = cvCreateImage(cvGetSize(imgGrey),IPL_DEPTH_8U,1);    cvCopyImage(srcImgGrey,imgBasicGlobalThreshold);int  pg[256],i,thre;    for (i=0;i<256;i++) pg[i]=0;for (i=0;i<imgBasicGlobalThreshold->imageSize;i++)      //  直方图统计         pg[(BYTE)imgBasicGlobalThreshold->imageData[i]]++;        thre = BasicGlobalThreshold(pg,0,256);    //  确定阈值     cout<<"The Threshold of this Image in BasicGlobalThreshold is:"<<thre<<endl;//输出显示阀值     cvThreshold(imgBasicGlobalThreshold,imgBasicGlobalThreshold,thre,255,CV_THRESH_BINARY);  //  二值化         cvNamedWindow("BasicGlobalThreshold", CV_WINDOW_AUTOSIZE );    cvShowImage( "BasicGlobalThreshold", imgBasicGlobalThreshold);//显示图像     cvReleaseImage(&imgBasicGlobalThreshold);/*---------------------------------------------------------------------------*//*OTSU*/    IplImage* imgOtsu = cvCreateImage(cvGetSize(imgGrey),IPL_DEPTH_8U,1);    cvCopyImage(srcImgGrey,imgOtsu);int thre2;    thre2 = otsu2(imgOtsu);    cout<<"The Threshold of this Image in Otsu is:"<<thre2<<endl;//输出显示阀值    cvThreshold(imgOtsu,imgOtsu,thre2,255,CV_THRESH_BINARY);  //  二值化        cvNamedWindow("imgOtsu", CV_WINDOW_AUTOSIZE );    cvShowImage( "imgOtsu", imgOtsu);//显示图像        cvReleaseImage(&imgOtsu);/*---------------------------------------------------------------------------*//*上下阀值法：利用正态分布求可信区间*/    IplImage* imgTopDown = cvCreateImage( cvGetSize(imgGrey), IPL_DEPTH_8U, 1 );    cvCopyImage(srcImgGrey,imgTopDown);    CvScalar mean ,std_dev;//平均值、 标准差double u_threshold,d_threshold;    cvAvgSdv(imgTopDown,&mean,&std_dev,NULL);        u_threshold = mean.val[0] +2.5* std_dev.val[0];//上阀值    d_threshold = mean.val[0] -2.5* std_dev.val[0];//下阀值//u_threshold = mean + 2.5 * std_dev; //错误//d_threshold = mean - 2.5 * std_dev;    cout<<"The TopThreshold of this Image in TopDown is:"<<d_threshold<<endl;//输出显示阀值    cout<<"The DownThreshold of this Image in TopDown is:"<<u_threshold<<endl;    cvThreshold(imgTopDown,imgTopDown,d_threshold,u_threshold,CV_THRESH_BINARY_INV);//上下阀值    cvNamedWindow("imgTopDown", CV_WINDOW_AUTOSIZE );    cvShowImage( "imgTopDown", imgTopDown);//显示图像        cvReleaseImage(&imgTopDown);/*---------------------------------------------------------------------------*//*迭代法*/    IplImage* imgIteration = cvCreateImage( cvGetSize(imgGrey), IPL_DEPTH_8U, 1 );    cvCopyImage(srcImgGrey,imgIteration);int thre3,nDiffRec;    thre3 =DetectThreshold(imgIteration, 100, nDiffRec);    cout<<"The Threshold of this Image in imgIteration is:"<<thre3<<endl;//输出显示阀值    cvThreshold(imgIteration,imgIteration,thre3,255,CV_THRESH_BINARY_INV);//上下阀值    cvNamedWindow("imgIteration", CV_WINDOW_AUTOSIZE );    cvShowImage( "imgIteration", imgIteration);    cvReleaseImage(&imgIteration);

Otsu代码一

代码/*======================================================================*//* OTSU global thresholding routine *//* takes a 2D unsigned char array pointer, number of rows, and *//* number of cols in the array. returns the value of the threshold *//*parameter: *image --- buffer for imagerows, cols --- size of imagex0, y0, dx, dy --- region of vector used for computing thresholdvvv --- debug option, is 0, no debug information outputed*//*OTSU 算法可以说是自适应计算单阈值（用来转换灰度图像为二值图像）的简单高效方法。下面的代码最早由 Ryan Dibble提供，此后经过多人Joerg.Schulenburg, R.Z.Liu 等修改，补正。算法对输入的灰度图像的直方图进行分析，将直方图分成两个部分，使得两部分之间的距离最大。划分点就是求得的阈值。*//*======================================================================*/int otsu (unsigned char*image, int rows, int cols, int x0, int y0, int dx, int dy, int vvv){        unsigned char*np; // 图像指针int thresholdValue=1; // 阈值int ihist[256]; // 图像直方图，256个点    int i, j, k; // various countersint n, n1, n2, gmin, gmax;double m1, m2, sum, csum, fmax, sb;    // 对直方图置零    memset(ihist, 0, sizeof(ihist));        gmin=255; gmax=0;// 生成直方图for (i = y0 +1; i < y0 + dy -1; i++)     {        np = (unsigned char*)image[i*cols+x0+1];for (j = x0 +1; j < x0 + dx -1; j++)        {            ihist[*np]++;if(*np > gmax) gmax=*np;if(*np < gmin) gmin=*np;            np++; /* next pixel */        }    }    // set up everything    sum = csum =0.0;    n =0;    for (k =0; k <=255; k++)     {        sum += (double) k * (double) ihist[k]; /* x*f(x) 质量矩*/        n += ihist[k]; /* f(x) 质量 */    }    if (!n)     {// if n has no value, there is problems...        fprintf (stderr, "NOT NORMAL thresholdValue = 160\n");return (160);    }    // do the otsu global thresholding method    fmax =-1.0;    n1 =0;for (k =0; k <255; k++)    {        n1 += ihist[k];if (!n1)         { continue;         }        n2 = n - n1;if (n2 ==0)        { break;         }        csum += (double) k *ihist[k];        m1 = csum / n1;        m2 = (sum - csum) / n2;        sb = (double) n1 *(double) n2 *(m1 - m2) * (m1 - m2);/* bbg: note: can be optimized. */if (sb > fmax)         {            fmax = sb;            thresholdValue = k;        }    }    // at this point we have our thresholding value    // debug code to display thresholding valuesif ( vvv &1 )        fprintf(stderr,"# OTSU: thresholdValue = %d gmin=%d gmax=%d\n",        thresholdValue, gmin, gmax);    return(thresholdValue);}

Otsu代码二 

代码/*======================================================================*//* OTSU global thresholding routine *//*======================================================================*/int otsu2 (IplImage *image){int w = image->width;int h = image->height;        unsigned char*np; // 图像指针    unsigned char pixel;int thresholdValue=1; // 阈值int ihist[256]; // 图像直方图，256个点    int i, j, k; // various countersint n, n1, n2, gmin, gmax;double m1, m2, sum, csum, fmax, sb;    // 对直方图置零...    memset(ihist, 0, sizeof(ihist));        gmin=255; gmax=0;// 生成直方图for (i =0; i < h; i++)     {        np = (unsigned char*)(image->imageData + image->widthStep*i);for (j =0; j < w; j++)         {            pixel = np[j];            ihist[ pixel]++;if(pixel > gmax) gmax= pixel;if(pixel < gmin) gmin= pixel;        }    }    // set up everything    sum = csum =0.0;    n =0;    for (k =0; k <=255; k++)     {        sum += k * ihist[k]; /* x*f(x) 质量矩*/        n += ihist[k]; /* f(x) 质量 */    }    if (!n)     {// if n has no value, there is problems...//fprintf (stderr, "NOT NORMAL thresholdValue = 160\n");        thresholdValue =160;goto L;    }    // do the otsu global thresholding method    fmax =-1.0;    n1 =0;for (k =0; k <255; k++)     {        n1 += ihist[k];if (!n1) { continue; }        n2 = n - n1;if (n2 ==0) { break; }        csum += k *ihist[k];        m1 = csum / n1;        m2 = (sum - csum) / n2;        sb = n1 * n2 *(m1 - m2) * (m1 - m2);/* bbg: note: can be optimized. */if (sb > fmax)        {            fmax = sb;            thresholdValue = k;        }    }    L:for (i =0; i < h; i++)     {        np = (unsigned char*)(image->imageData + image->widthStep*i);for (j =0; j < w; j++)         {if(np[j] >= thresholdValue)                np[j] =255;else np[j] =0;        }    }//cout<<"The Threshold of this Image in Otsu is:"<<thresholdValue<<endl;return(thresholdValue);}

最大熵阀值

代码/*=============================================================================  代码内容：最大熵阈值分割                                      =  修改日期:2009-3-3                                                                                                         =  作者:crond123 =  博客:http://blog.csdn.net/crond123/=  E_Mail:[e-mail]                                                      ===============================================================================*/// 计算当前位置的能量熵double caculateCurrentEntropy(CvHistogram * Histogram1,int cur_threshold,entropy_state state){int start,end;int  total =0;double cur_entropy =0.0;if(state == back)        {        start =0;        end = cur_threshold;        }else        {        start = cur_threshold;        end =256;        }    for(int i=start;i<end;i++)        {        total += (int)cvQueryHistValue_1D(Histogram1,i);//查询直方块的值 P304    }for(int j=start;j<end;j++)    {if((int)cvQueryHistValue_1D(Histogram1,j)==0)continue;double percentage = cvQueryHistValue_1D(Histogram1,j)/total;/*熵的定义公式*/        cur_entropy +=-percentage*logf(percentage);/*根据泰勒展式去掉高次项得到的熵的近似计算公式        cur_entropy += percentage*percentage;*/            }return cur_entropy;//    return (1-cur_entropy);}//寻找最大熵阈值并分割void  MaxEntropy(IplImage *src,IplImage *dst){    assert(src != NULL);    assert(src->depth ==8&& dst->depth ==8);    assert(src->nChannels ==1);    CvHistogram * hist  = cvCreateHist(1,&HistogramBins,CV_HIST_ARRAY,HistogramRange);//创建一个指定尺寸的直方图//参数含义：直方图包含的维数、直方图维数尺寸的数组、直方图的表示格式、方块范围数组、归一化标志    cvCalcHist(&src,hist);//计算直方图double maxentropy =-1.0;int max_index =-1;// 循环测试每个分割点，寻找到最大的阈值分割点for(int i=0;i<HistogramBins;i++)        {double cur_entropy = caculateCurrentEntropy(hist,i,object)+caculateCurrentEntropy(hist,i,back);if(cur_entropy>maxentropy)        {            maxentropy = cur_entropy;            max_index = i;        }    }    cout<<"The Threshold of this Image in MaxEntropy is:"<<max_index<<endl;    cvThreshold(src, dst, (double)max_index,255, CV_THRESH_BINARY);    cvReleaseHist(&hist);}

基本全局阀值法

代码
/*============================================================================
=  代码内容：基本全局阈值法                             
==============================================================================*/

代码/*=============================================================================  代码内容：基本全局阈值法                              ==============================================================================*/int BasicGlobalThreshold(int*pg,int start,int end){                                           //  基本全局阈值法int  i,t,t1,t2,k1,k2;double u,u1,u2;        t=0;         u=0;for (i=start;i<end;i++)     {        t+=pg[i];                u+=i*pg[i];    }    k2=(int) (u/t);                          //  计算此范围灰度的平均值    do     {        k1=k2;        t1=0;            u1=0;for (i=start;i<=k1;i++)         {             //  计算低灰度组的累加和            t1+=pg[i];                u1+=i*pg[i];        }        t2=t-t1;        u2=u-u1;if (t1)             u1=u1/t1;                     //  计算低灰度组的平均值else             u1=0;if (t2)             u2=u2/t2;                     //  计算高灰度组的平均值else             u2=0;        k2=(int) ((u1+u2)/2);                 //  得到新的阈值估计值    }while(k1!=k2);                           //  数据未稳定，继续//cout<<"The Threshold of this Image in BasicGlobalThreshold is:"<<k1<<endl;return(k1);                              //  返回阈值}