txtrsubs.c

This is code tutorial for image processing include:histogram,sketon....

    /********************************************
    *
    *    file txtrsubs.c
    *
    *    Functions: This file contains
    *       sigma
    *       skewness
    *       amean
    *       adifference
    *       difference_array
    *       hurst
    *       compare
    *       get_texture_options
    *
    *    Purpose:
    *       These functions calculate measures
    *       that help distinguish textures.
    *
    *    External Calls:
    *       utility.c - fix_edges
    *                   sort_elements
    *       fitt.c - fit
    *
    *    Modifications:
    *       12 August 1993- created
    *       27 August 1998 - modified to work on 
    *            entire images at once.
    *
    ********************************************/

#include "cips.h"



     /*******************************************
     *
     *   sigma(..
     *
     *   This calculates the variance and the 
     *   sigma for a sizeXsize area.
     *
     *   It sums the squares of the difference
     *   between each pixel and the mean of
     *   the area and divides that by the
     *   number of pixels in the area.
     *
     *   The output image is set to the square
     *   root of the variance since the variance
     *   will almost certainly be out of range
     *   for the image.  The square root of the
     *   variance will be sigma.
     *
     *******************************************/

sigma(the_image, out_image,
      size, threshold, high,
      rows, cols, bits_per_pixel)
   int    high, threshold, size;
   long   bits_per_pixel, cols, rows;
   short  **the_image,
          **out_image;
{
   int      a, b, count, i, j, k,
            max, mean, new_hi, new_low,
            sd2, sd2p1;
   short    sigma;
   unsigned long diff, variance;

   sd2   = size/2;
   sd2p1 = sd2 + 1;

   max     = 255;
   new_hi  = 250;
   new_low = 16;
   if(bits_per_pixel == 4){
       new_hi  = 10;
       new_low = 3;
       max     = 16;
   }

      /***************************
      *
      *   Loop over image array
      *
      ****************************/

   printf("\n");
   for(i=sd2; i<rows-sd2; i++){
      if( (i%10) == 0) printf("%d ", i);
      for(j=sd2; j<cols-sd2; j++){

            /*****************************
            *
            *   Run through the small area
            *   and calculate the mean.
            *
            ******************************/

         mean = 0;
         for(a=-sd2; a<sd2p1; a++){
            for(b=-sd2; b<sd2p1; b++){
               mean = mean + the_image[i+a][j+b];
            }
         }
         mean = mean/(size*size);

            /*****************************
            *
            *   Run through the small area
            *   again and the calculate the
            *   variance.
            *
            ******************************/

         variance = 0;
         diff     = 0;
         for(a=-sd2; a<sd2p1; a++){
            for(b=-sd2; b<sd2p1; b++){
               diff     = the_image[i+a][j+b] - mean;
               variance = variance + (diff*diff);
            }
         }

         variance = variance/(size*size);
         sigma    = sqrt(variance);
         if(sigma > max) sigma = max;
         out_image[i][j] = sigma;

      }  /* ends loop over j */
   }  /* ends loop over i */


     /* if desired, threshold the output image */
   if(threshold == 1){
       for(i=0; i<rows; i++){
          for(j=0; j<cols; j++){
             if(out_image[i][j] > high){
                  out_image[i][j] = new_hi;
             }
             else{
                  out_image[i][j] = new_low;
             }
          }
       }
   }  /* ends if threshold == 1 */

   fix_edges(out_image, sd2, rows-1, cols-1);

}  /* ends sigma */





     /*******************************************
     *
     *   skewness(..
     *
     *   This calculates the skewness for a
     *   sizeXsize area.
     *
     *   Look at Levine's book page 449 for
     *   the formula.
     *   "Vision in Man and Machine" by
     *   Martin D. Levine, McGraw Hill, 1985.
     *
     *******************************************/

skewness(the_image, out_image,
         size, threshold, high,
         rows, cols, bits_per_pixel)
   int    high, threshold, size;
   long   bits_per_pixel, cols, rows;
   short  **the_image,
          **out_image;
{
   int      a, b, count, i, j, k,
            max, mean, new_hi, new_low,
            sd2, sd2p1;
   long     cube;
   short    sigma, skew;
   unsigned long diff, sigma3, variance;

   sd2   = size/2;
   sd2p1 = sd2 + 1;

   max     = 255;
   new_hi  = 250;
   new_low = 16;
   if(bits_per_pixel == 4){
       new_hi  = 10;
       new_low = 3;
       max     = 16;
   }

      /***************************
      *
      *   Loop over image array
      *
      ****************************/

   printf("\n");
   for(i=sd2; i<rows-sd2; i++){
      if( (i%10) == 0) printf("%d ", i);
      for(j=sd2; j<cols-sd2; j++){

            /*****************************
            *
            *   Run through the small area
            *   and calculate the mean.
            *
            ******************************/

         mean = 0;
         for(a=-sd2; a<sd2p1; a++){
            for(b=-sd2; b<sd2p1; b++){
               mean = mean + the_image[i+a][j+b];
            }
         }
         mean = mean/(size*size);

            /*****************************
            *
            *   Run through the small area
            *   again and the calculate the
            *   variance and the cube.
            *
            ******************************/

         variance = 0;
         diff     = 0;
         cube     = 0;
         for(a=-sd2; a<sd2p1; a++){
            for(b=-sd2; b<sd2p1; b++){
               diff     = the_image[i+a][j+b] - mean;
               cube     = cube + (diff*diff*diff);
               variance = variance + (diff*diff);
            }
         }

         variance        = variance/(size*size);
         sigma           = sqrt(variance);
         sigma3          = sigma*sigma*sigma;
         if(sigma3 == 0)
            sigma3       = 1;
         skew            = cube/(sigma3*size*size);
         out_image[i][j] = skew;
         if(out_image[i][j] > max)
            out_image[i][j] = max;

      }  /* ends loop over j */
   }  /* ends loop over i */



     /* if desired, threshold the output image */
   if(threshold == 1){
       for(i=0; i<rows; i++){
          for(j=0; j<cols; j++){
             if(out_image[i][j] > high){
                  out_image[i][j] = new_hi;
             }
             else{
                  out_image[i][j] = new_low;
             }
          }
       }
   }  /* ends if threshold == 1 */

   fix_edges(out_image, sd2, rows-1, cols-1);

}  /* ends skewness */




     /*******************************************
     *
     *   adifference(..
     *
     *   This function performs the difference
     *   operation for a specified array
     *   in an image file.
     *
     *******************************************/

adifference(the_image, out_image,
            size,
            rows, cols)
   int    size;
   long   cols, rows;
   short  **the_image,
          **out_image;
{
   int      sd2, sd2p1;

   sd2   = size/2;
   sd2p1 = sd2 + 1;

   difference_array(the_image, out_image, 
                    size, rows, cols);

   fix_edges(out_image, sd2, rows-1, cols-1);

}  /* ends adifference */





     /*******************************************
     *
     *   difference_array(..
     *
     *   This function takes the input image
     *   array the_image and places in out_image
     *   the gray level differences of the pixels
     *   in the_image.  It uses the size
     *   parameter for the distance between pixels
     *   used to get the difference.
     *
     *******************************************/

difference_array(the_image, out_image, 
                 size, rows, cols)
   int    size;
   long   cols, rows;
   short  **the_image,
          **out_image;
{
   int i, j, sd2;
   sd2   = size/2;

   for(i=sd2; i<rows-sd2; i++){
      for(j=sd2; j<cols-sd2; j++){
         out_image[i][j] =
            abs(the_image[i][j] - 
                the_image[i+sd2][j+sd2]);
      }  /* ends loop over j */
   }  /* ends loop over i */

   fix_edges(out_image, sd2, rows-1, cols-1);

}  /* ends difference_array */




     /*******************************************
     *
     *   amean(..
     *
     *   This calculates the mean measure
     *   for a sizeXsize area.
     *
     *   Look at Levine's book page 451 for
     *   the formula.
     *   "Vision in Man and Machine" by
     *   Martin D. Levine, McGraw Hill, 1985.
     *
     *******************************************/

amean(the_image, out_image,
      size,
      rows, cols, bits_per_pixel)
   int    size;
   long   bits_per_pixel, cols, rows;
   short  **the_image,
          **out_image;
{
   int      a, b, count, i, j, k, max,
            sd2, sd2p1;
   short    pixel;
   unsigned long big;

   sd2   = size/2;
   sd2p1 = sd2 + 1;

   max = 255;
   if(bits_per_pixel == 4){
       max = 16;
   }

      /**************************************
      *
      *   Calculate the gray level difference
      *   array.
      *
      ***************************************/

   difference_array(the_image, out_image, 
                    size, rows, cols);
   for(i=0; i<rows; i++)
      for(j=0; j<cols; j++)
         the_image[i][j] = out_image[i][j];

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