SAD with full search range [-6,6]

I have an image divided in 16x16 macroblocks and another image 144x176 and I want to calculate SAD for each macroblock in range of [-6,6] for vectors (i,j) using full search

lets assume m=144 and n=176 so my macroblocks cell has dimensions 144/16 176/16 = 9x11 I want to take the sum of absolute differences from each macroblock and the 144x176 image. The problem is that I go out of limits of my image array or I go till the end of the rows and columns instead I only need the values that are +-6.

Any idea ?

for i =1:1:9    %i take each 16x16 macroblock 
  for j =1:1:11
    for k=6:1:m  
      for l=6:1:n
          kl=kl+sum(abs(macroblocks{i,j} - image(k-5:k+5,l-5:l+5)));
      end
    end
  end
end

Answer 1

[Image Analyst]

Try blockproc(). I attach demos where I use it for other functions.

[Lefteris Kostakis]

I have already divided my first image using blockproc now I wanna make the sad algorithm

[Image Analyst]

Sorry, I didn't realize that since I didn't see your call to blockproc(). So I think the way I'd proceed is to use imresize() with the 'nearest' option to make the small matrix the same size as your big matrix. Then compute the SAD

diffImage = double(image1) - double(image2); % Must cast integer images to double to avoid clipping of integers
imshow(diffImage, []);
SAD = sum(abs(diffImage(:)))

[Lefteris Kostakis]

I want to make it with for loops as I want to extract some results

macroblock(x,y) - (image(x,y) for (i,j)=[-6,6])

[Image Analyst]

OK, so do it with for loops but I think it will make the computation of the row and column indexes easier if you use imresize() to make them the same size.

[Lefteris Kostakis]

thats why i posted, to get help doing it with for loops :P