原始的Bernsen算法

最新推荐文章于 2024-04-22 08:05:32 发布

weixin_30549657

最新推荐文章于 2024-04-22 08:05:32 发布

阅读量299

点赞数

文章标签： python

原文链接：http://www.cnblogs.com/lingde27/archive/2011/04/29/2032759.html

版权

较原始的Bernsen：

这个算法的中心思想是：设当前像素为P，计算以P为中心的大小为(2w+1)*(2w+1)窗口内的所有像素的最大值M与最小值N，两者的均值T,

if(M-N）> S

则当前点P的阈值为T。

else

当前窗口所在区域的灰度级差别较小，那么窗口在目标区或在背景区，若T>T'则当前点灰度值为255，否则，当前点灰度值为0.

S作者最初设为15, T'设为（255+0）/2=128。

这种最原始的算法的效果总体来说还行，但一般所用的Bernsen算法都是经过各种改进的。

#define Th 128
void Bernsen(IplImage *src, IplImage *dst)
{
int i = 0, j = 0,k = 0, l = 0, maximum = 0, minor = 0, mid_gray = 0;
int upleft = 0, upright = 0, downleft = 0, downright = 0;
int wide = src->widthStep;
int high = src->height;
unsigned long sum = 0;
int average = 0;
unsigned char *p, *q;
p = (unsigned char *)src->imageData;
q = (unsigned char *)dst->imageData;
for (j = 0; j < high; j++, p += wide)
{
  for (i = 0; i < wide; i++)
  {
   sum += p[i];
  }
}
average = sum / (wide * high);
for (j = 0; j < high; j++)
{
  p = (unsigned char *)(src->imageData + j * wide);
  q = (unsigned char *)(dst->imageData + j * wide);
  for (i = 0; i < wide; i++)
  {
   upleft = (i - 15) > 0 ? (i-15):0;
   upright = (i + 15) < wide ? (i+15):(wide- 1);
   downleft = (j - 15) > 0 ? (j - 15):0;
   downright = (j + 15) < high ? (j + 15):(high - 1);
   for (k = downright; k <= downright; k++)
   {
    for (l = upleft; l <= upright; l++)
    {
     if (p[l] >= maximum)
     {
      maximum = p[l];
     }
      else if (p[l] <= minor)
      {
       minor = p[l];
      }
    }
   }
   mid_gray = (maximum + minor) >> 1;
   if ((maximum - minor) > S)
   {
    q[i] = p[i] >= mid_gray ? 255 : 0;
   }
   else
    q[i] = mid_gray >= average ? 255 : 0;
  }
}
}