简单的图像细化算法

*   函数名称：

 *   Tinning()

 *    参数:

 *   int  Width               源图像宽度（象素数）

 *   int  Height               源图像高度（象素数）

 *    int  TransImage[]                源图象

 *   说明:

 *   该函数利用基于索引表的细化算法对图像进行细化。

 ***********************************************************/

#define TRUE 1
#define FALSE 0
typedef unsigned char BYTE;

void Thinning(int Width,int Height,int TransImage[])
{
 int i,j;
 int num;
 bool Finished;    // 细化结束标志
// int TransImage[1000];
 // 各个变量用来存储(i,j)位置的八邻域像素点的灰度
 int nw,n,ne,w,e,sw,s,se;
 static int erasetable[256]={   // 细化表
   0,0,1,1,0,0,1,1,
   1,1,0,1,1,1,0,1,
   1,1,0,0,1,1,1,1,
   0,0,0,0,0,0,0,1,
   
   0,0,1,1,0,0,1,1,
   1,1,0,1,1,1,0,1,
   1,1,0,0,1,1,1,1,
   0,0,0,0,0,0,0,1,
   
   1,1,0,0,1,1,0,0,
   0,0,0,0,0,0,0,0,
   0,0,0,0,0,0,0,0,
   0,0,0,0,0,0,0,0,
   
   1,1,0,0,1,1,0,0,
   1,1,0,1,1,1,0,1,
   0,0,0,0,0,0,0,0,
   0,0,0,0,0,0,0,0,
   
   0,0,1,1,0,0,1,1,
   1,1,0,1,1,1,0,1,
   1,1,0,0,1,1,1,1,
   0,0,0,0,0,0,0,1,
   
   0,0,1,1,0,0,1,1,
   1,1,0,1,1,1,0,1,
   1,1,0,0,1,1,1,1,
   0,0,0,0,0,0,0,0,
   
   1,1,0,0,1,1,0,0,
   0,0,0,0,0,0,0,0,
   1,1,0,0,1,1,1,1,
   0,0,0,0,0,0,0,0,
   
   1,1,0,0,1,1,0,0,
   1,1,0,1,1,1,0,0,
   1,1,0,0,1,1,1,0,
   1,1,0,0,1,0,0,0
 };
 Finished=FALSE;
 // 在细化之前将图像边缘去掉，防止干扰
 for(i=0;i<Width;i++){
  for(j=0;j<20;j++){
   TransImage[j*Width+i]=255;
   TransImage[(Height-j)*Width+i]=255;
  }
 }
 for(j=0;j<Height;j++){
  for(i=0;i<15;i++){
   TransImage[j*Width+i]=255;
   TransImage[j*Width+(Width-i-1)]=255;
  }
 }
 // 开始细化
 while(!Finished){
  Finished=TRUE;
  // 水平扫描
  for (j=0;j<Height;j++){
   for(i=0;i<Width;i++){
    if(TransImage[Width*(Height-j-1)+i]==0){
     w=TransImage[Width*(Height-j-1)+i-1];
     e=TransImage[Width*(Height-j-1)+i+1];
     // 判断(i,j)是否是边界点，如是，求该点的八邻域灰度值(0/255)，
// 根据各点的权重，计算对应查找表的索引；
     if( (w==255)|| (e==255)){
      nw=TransImage[Width*(Height-j)+i-1];
      n= TransImage[Width*(Height-j)+i];
      ne=TransImage[Width*(Height-j)+i+1];
      sw=TransImage[Width*(Height-j-2)+i-1];
      s= TransImage[Width*(Height-j-2)+i];
      se=TransImage[Width*(Height-j-2)+i+1];

      num = nw/255+n/255*2+ne/255*4+w/255*8+e/255*16+w/255*32+s/255*64+se/255*128;
      if(erasetable[num]==1){
       // 查表，如果符合条件，将边界点修改为图像的背景
       TransImage[Width*(Height-j-1)+i]=(BYTE)255;
       Finished=FALSE;     // 再次进行扫描
       i++;
      }
     }
    }
   }
  }
  // 垂直扫描
  for (i=0;i<Width;i++){
   for(j=0;j<Height;j++){
    if(TransImage[Width*(Height-j-1)+i]==0){
     n=TransImage[Width*(Height-j)+i];
     s=TransImage[Width*(Height-j-2)+i];
     if( (n==255)|| (s==255)){
      nw=TransImage[Width*(Height-j)+i-1];
      ne=TransImage[Width*(Height-j)+i+1];
      w= TransImage[Width*(Height-j-1)+i-1];
      e= TransImage[Width*(Height-j-1)+i+1];
      sw=TransImage[Width*(Height-j-2)+i-1];
      se=TransImage[Width*(Height-j-2)+i+1];
      num = nw/255+n/255*2+ne/255*4+w/255*8+e/255*16+sw/255*32+s/255*64+se/255*128;
      if(erasetable[num]==1){
       // 查表，如果符合条件，将边界点修改为图像的背景
       TransImage[Width*(Height-j-1)+i]=(BYTE)255;
       Finished=FALSE;     // 再次进行扫描
       j++;
      }
     }
    }
   }
  } // 垂直扫描End
 } // 细化End
 return;
}