OTSU方法计算图像二值化的自适应阈值

/*
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。

转自：http://forum.assuredigit.com/display_topic_threads.asp?ForumID=8&TopicID=3480
 
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。
  
 parameter:   *image              --- buffer for image
                      rows, cols        --- size of image
                      x0, y0, dx, dy   --- region of vector used for computing threshold
                      vvv                  --- debug option, is 0, no debug information outputed
 */
/*======================================================================*/
/*   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       */
/*======================================================================*/
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 counters
  int 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 = &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 values
  if ( vvv & 1 )
  fprintf(stderr,"# OTSU: thresholdValue = %d gmin=%d gmax=%d\n",
     thresholdValue, gmin, gmax);

  return(thresholdValue);
}

本文地址：http://com.8s8s.com/it/it27930.htm