Hello,

I am working on a program which will apply a median filter to an image then save and load the result, then do this again repeatedly.  My program is functional though the results are less than ideal because I am applying a 3x3 matrix to collect the values for one pixel's nearest neighbors and in order to keep the array from going out of bounds I am neglecting the first and last columns and rows.  The results for this leave those pixels black which then alters the next iteration to where eventually I am left with a solid black frame.

To sum up, I would like to know how to handle these border pixels so that the returned image is a consistent alteration of the original.

Any help, advice, or direction is much appreciated.

Thank you,
Matthew

1. /**
    *    Matthew Lord : 2010 / 11 . 30
    *
    *    using a variety of examples from:
    *        "Processing" by Fry & Reas
    *
    *    This code will process an image
    *
    */
   
   import java.util.Arrays;
   
   PImage img;
   
   int matrixSize = 9;
   int rgbListMedian = 4;
   int offset = 1;
   
   
   // SETUP
   void setup( ) {
       size( 320, 180 );
   }
   
   
   // DRAW
   void draw( ) {
   
       // to render the first "n" images
       if( frameCount <= 620 ) {
      
           // select an image
           if( frameCount <= 1 ) {
               // load the seed image
               img = loadImage( "testimage.jpg" );
           } else if( frameCount > 1 ) {
               // assign the new image to the "img" variable
               img = loadImage( "processedimgs/testimage." + nf( frameCount - 1, 4 ) + ".tif" );
           }
          
           // load the pixels of the image
           img.loadPixels( );
          
           // Create a new image to store the new data into
           // Create an opaque image of the same size as the original
           PImage medianImg = createImage( img.width, img.height, RGB );
      
           // cycle through all pixels
           // Loop through every pixel in the image.
           for( int y = offset; y < img.height - offset; y++ ) {
          
               for( int x = offset; x < img.width - offset; x++ ) {
              
                   int z = 0;
                  
                   int[ ] AList = new int[ matrixSize ];
                   int[ ] RList = new int[ matrixSize ];
                   int[ ] GList = new int[ matrixSize ];
                   int[ ] BList = new int[ matrixSize ];
                  
                   for( int ky = -offset; ky <= offset; ky++ ) {
                  
                       for( int kx = -offset; kx <= offset; kx++ ) {
                      
                           int pos = ( y + ky ) * width + ( x + kx );
                          
                           int alphaVal = int( alpha( img.pixels[ pos ] ) );
                           int redVal = int( red( img.pixels[ pos ] ) );
                           int greenVal = int( green( img.pixels[ pos ] ) );
                           int blueVal = int( blue( img.pixels[ pos ] ) );
                          
                           // place the RGB values into their respective arrays
                           AList[ z ] = int( alphaVal );
                           RList[ z ] = int( redVal );
                           GList[ z ] = int( greenVal );
                           BList[ z ] = int( blueVal );
                          
                           z++;
                       }
                  
                   }
                                  
                   // sort the Arrays
                   Arrays.sort( AList );
                   Arrays.sort( RList );
                   Arrays.sort( GList );
                   Arrays.sort( BList );
                  
                   // select the median value for each array
                   color medianColor = color( RList[ rgbListMedian - 1 ], GList[ rgbListMedian - 1 ], BList[ rgbListMedian - 1 ], AList[ rgbListMedian - 1 ] );
                  
                   // assign the RGB values to the new image pixel in the same location as the original
                   medianImg.pixels[ y * img.width + x ] = color( medianColor );
              
               }
          
           }
          
           medianImg.updatePixels( );
           image( medianImg, 0, 0 );
          
           saveFrame( "processedimgs/testimage.####.tif" );
      
       }
              
   }