opencv实现二值图像细化的算法

转自：http://blog.csdn.net/byxdaz/archive/2010/06/02/5642669.aspx

细化算法通常和骨骼化、骨架化算法是相同的意思，也就是thin算法或者skeleton算法。虽然很多图像处理的教材上不是这么写的，具体原因可以看这篇论文，Louisa Lam, Seong-Whan Lee, Ching Y. Suen，“Thinning Methodologies-A Comprehensive Survey ”，IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 14, NO. 9, SEPTEMBER 1992 ，总结了几乎所有92年以前的经典细化算法。

函数：void cvThin( IplImage* src, IplImage* dst, int iterations=1)
功能：将IPL_DEPTH_8U型二值图像进行细化
参数：src，原始IPL_DEPTH_8U型二值图像
dst，目标存储空间，必须事先分配好，且和原图像大小类型一致
iterations，迭代次数
参考文献：T. Y. Zhang and C. Y. Suen, “A fast parallel algorithm for thinning digital patterns,” Comm. ACM, vol. 27, no. 3, pp. 236-239, 1984.

void cvThin( IplImage* src, IplImage* dst, int iterations=1)
{
CvSize size = cvGetSize(src);

cvCopy(src, dst);
    int n = 0,i = 0,j = 0;
for(n=0; n<iterations; n++)
{
IplImage* t_image = cvCloneImage(dst);
for(i=0; i<size.height; i++)
{
   for(j=0; j<size.width; j++)
   {
    if(CV_IMAGE_ELEM(t_image,byte,i,j)==1)
    {
     int ap=0;
     int p2 = (i==0)?0:CV_IMAGE_ELEM(t_image,byte, i-1, j);
     int p3 = (i==0 || j==size.width-1)?0:CV_IMAGE_ELEM(t_image,byte, i-1, j+1);
     if (p2==0 && p3==1)
     {
      ap++;
     }
     int p4 = (j==size.width-1)?0:CV_IMAGE_ELEM(t_image,byte,i,j+1);
     if(p3==0 && p4==1)
     {
      ap++;
     }
     int p5 = (i==size.height-1 || j==size.width-1)?0:CV_IMAGE_ELEM(t_image,byte,i+1,j+1);
     if(p4==0 && p5==1)
     {
      ap++;
     }
     int p6 = (i==size.height-1)?0:CV_IMAGE_ELEM(t_image,byte,i+1,j);
     if(p5==0 && p6==1)
     {
      ap++;
     }
     int p7 = (i==size.height-1 || j==0)?0:CV_IMAGE_ELEM(t_image,byte,i+1,j-1);
     if(p6==0 && p7==1)
     {
      ap++;
     }
     int p8 = (j==0)?0:CV_IMAGE_ELEM(t_image,byte,i,j-1);
     if(p7==0 && p8==1)
     {
      ap++;
     }
     int p9 = (i==0 || j==0)?0:CV_IMAGE_ELEM(t_image,byte,i-1,j-1);
     if(p8==0 && p9==1)
     {
      ap++;
     }
     if(p9==0 && p2==1)
     {
      ap++;
     }
     if((p2+p3+p4+p5+p6+p7+p8+p9)>1 && (p2+p3+p4+p5+p6+p7+p8+p9)<7)
     {
      if(ap==1)
      {
       if(!(p2 && p4 && p6))
       {
        if(!(p4 && p6 && p8))
        {
         CV_IMAGE_ELEM(dst,byte,i,j)=0;
        }
       }
      }
     }

}
   }
}
cvReleaseImage(&t_image);
t_image = cvCloneImage(dst);
for(i=0; i<size.height; i++)
{
   for(int j=0; j<size.width; j++)
   {
    if(CV_IMAGE_ELEM(t_image,byte,i,j)==1)
    {
     int ap=0;
     int p2 = (i==0)?0:CV_IMAGE_ELEM(t_image,byte, i-1, j);
     int p3 = (i==0 || j==size.width-1)?0:CV_IMAGE_ELEM(t_image,byte, i-1, j+1);
     if (p2==0 && p3==1)
     {
      ap++;
     }
     int p4 = (j==size.width-1)?0:CV_IMAGE_ELEM(t_image,byte,i,j+1);
     if(p3==0 && p4==1)
     {
      ap++;
     }
     int p5 = (i==size.height-1 || j==size.width-1)?0:CV_IMAGE_ELEM(t_image,byte,i+1,j+1);
     if(p4==0 && p5==1)
     {
      ap++;
     }
     int p6 = (i==size.height-1)?0:CV_IMAGE_ELEM(t_image,byte,i+1,j);
     if(p5==0 && p6==1)
     {
      ap++;
     }
     int p7 = (i==size.height-1 || j==0)?0:CV_IMAGE_ELEM(t_image,byte,i+1,j-1);
     if(p6==0 && p7==1)
     {
      ap++;
     }
     int p8 = (j==0)?0:CV_IMAGE_ELEM(t_image,byte,i,j-1);
     if(p7==0 && p8==1)
     {
      ap++;
     }
     int p9 = (i==0 || j==0)?0:CV_IMAGE_ELEM(t_image,byte,i-1,j-1);
     if(p8==0 && p9==1)
     {
      ap++;
     }
     if(p9==0 && p2==1)
     {
      ap++;
     }
     if((p2+p3+p4+p5+p6+p7+p8+p9)>1 && (p2+p3+p4+p5+p6+p7+p8+p9)<7)
     {
      if(ap==1)
      {
       if(p2*p4*p8==0)
       {
        if(p2*p6*p8==0)
        {
         CV_IMAGE_ELEM(dst, byte,i,j)=0;
        }
       }
      }
     }
    }

}

}
cvReleaseImage(&t_image);
}

}

//使用举例

#include "cxcore.h"
#include "cv.h"
#include "highgui.h"

int main(int argc, char* argv[])
{
if(argc!=2)
{
return 0;
}
IplImage *pSrc = NULL,*pDst = NULL,*pTmp = NULL;

//传入一个灰度图像
pSrc = cvLoadImage(argv[1],CV_LOAD_IMAGE_GRAYSCALE);
if(!pSrc)
{
return 0;
}
pTmp = cvCloneImage(pSrc);
    pDst = cvCreateImage(cvGetSize(pSrc),pSrc->depth,pSrc->nChannels);
cvZero(pDst);
cvThreshold(pSrc,pTmp,128,1,CV_THRESH_BINARY_INV);//做二值处理，将图像转换成0，1格式
//cvSaveImage("c://Threshold.bmp",pTmp,0);
cvThin(pTmp,pDst,8);//细化，通过修改iterations参数进一步细化
cvNamedWindow("src",1);
cvNamedWindow("dst",1);
cvShowImage("src",pSrc);
//将二值图像转换成灰度，以便显示
int i = 0,j = 0;
CvSize size = cvGetSize(pDst);
for(i=0; i<size.height; i++)
{
for(j=0; j<size.width; j++)
{
   if(CV_IMAGE_ELEM(pDst,uchar,i,j)==1)
   {
    CV_IMAGE_ELEM(pDst,uchar,i,j) = 0;
   }
   else
   {
    CV_IMAGE_ELEM(pDst,uchar,i,j) = 255;
   }
}
}
//cvSaveImage("c://thin.bmp",pDst);
cvShowImage("dst",pDst);
cvWaitKey(0);
    cvReleaseImage(&pSrc);
cvReleaseImage(&pDst);
cvReleaseImage(&pTmp);
cvDestroyWindow("src");
cvDestroyWindow("dst");
return 0;
}

opencv实现二值图像细化的算法相关推荐

OpenCV学习(14) 细化算法(2)
前面一篇教程中,我们实现了Zhang的快速并行细化算法,从算法原理上,我们可以知道,算法是基于像素8邻域的形状来决定是否删除当前像素.还有很多与此算法相似的细化算法,只是判断的条件不一样.在综述文章, ...
【opencv】图像细化
[opencv]图像细化 [opencv]图像细化 2014-02-17 21:03 5404人阅读评论(14) 收藏举报分类: opencv(1) 版权声明:本文为博主原创文章,未经博主允 ...
使用Python，OpenCV的Meanshift 和 Camshift 算法来查找和跟踪视频中的对象
使用Python,OpenCV的Meanshift 和 Camshift 算法来查找和跟踪视频中的对象 1. 效果图 2. 源码 2.1 MeanShift 2.2 Camshift(Continuo ...
OpenCV中直方图反向投影算法详解与实现
点击上方"小白学视觉",选择加"星标"或"置顶" 重磅干货,第一时间送达本文转自:opencv学堂一:直方图交叉 OpenCV中直方图反向 ...
OpenCV均值漂移meanshift algorithm算法的实例(附完整代码)
OpenCV均值漂移meanshift algorithm算法的实例 OpenCV均值漂移meanshift algorithm算法的实例 OpenCV均值漂移meanshift algorithm算 ...
OpenCV支持向量机SVM和SDG算法的实例(附完整代码)
OpenCV支持向量机SVM和SDG算法的实例 OpenCV支持向量机SVM和SDG算法的实例 OpenCV支持向量机SVM和SDG算法的实例 #include "opencv2/core. ...
OpenCV实现图像对齐ECC算法(附完整代码)
OpenCV实现图像对齐ECC算法 OpenCV实现图像对齐ECC算法 OpenCV实现图像对齐ECC算法 #include <opencv2/imgcodecs.hpp> #includ ...
OpenCV实现最大最小距离聚类算法
OpenCV实现最大最小距离聚类算法 [尊重原创,转载请注明出处]https://blog.csdn.net/guyuealian/article/details/80255524 本博客提供多版本的 ...
Win8 Metro(C#)数字图像处理--2.40二值图像轮廓提取算法
Win8 Metro(C#)数字图像处理--2.40二值图像轮廓提取算法原文:Win8 Metro(C#)数字图像处理--2.40二值图像轮廓提取算法 [函数名称] 二值图像轮廓提取 ...

opencv实现二值图像细化的算法

opencv实现二值图像细化的算法相关推荐

最新文章

热门文章