opencv实现二值图像细化的算法

opencv实现二值图像细化的算法

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;
}