MATLAB: How can i make this code faster? I am new here and generally in programing

disparity mapMATLAB

% define window size
prompt = {'Enter window size:'};
dlg_title = 'Input';
num_lines = 1;
defaultans = {'9'};
answer = inputdlg(prompt,dlg_title,num_lines,defaultans);
windowSize=str2double(answer(1,1));
%read the left and right images
Im_left=imread('im2.png');
Im_right=imread('im6.png');
%make them black and white
Im_left=double(rgb2gray(Im_left));
Im_right=double(rgb2gray(Im_right));
% find the size of the image
[rows,columes]=size(Im_left);
% create an empy matrix same size to image.
dispmap = zeros((rows), (columes));
% go through all rows of the image. starting from the row that is half
% window size away from the first row and ends to the one that is half
% window size away from the last row
for i=round(windowSize/2):rows-round(windowSize/2)
   % do the same for all columes
   for j=round(windowSize/2):columes-round(windowSize/2)
      % create an window that goes through the hole image.
      Window=Im_right(i-round(windowSize/2)+1:i+round(windowSize/2)-1,j-round(windowSize/2)+1:j+round(windowSize/2)-1);
      maxcost=inf;
      % go through the colume defined earlier up to 60 columes to the
      % right of this colume.
      if j+59+round(windowSize/2)<columes
         for  k=j:j+60
            % create another window that for each place of the first window
            % will go up to 60 columes to the right of the colume of the
            % first window
            Window_Try=Im_left(i-round(windowSize/2)+1:i+round(windowSize/2)-1,k-round(windowSize/2)+1:k+round(windowSize/2)-1);
              % calculate cost between the 2 windows in every step of second
              % window using SAD method
              %cost=sum(sum(abs(Window_Try-Window)));
              % calculate cost usind SSD method
              cost=sum(sum((Window_Try-Window).^2));
              % find the minumun cost for each central window of the first
              % image compeared with the 60 windows of the second image
              if cost<maxcost
                 maxcost=cost;
                 % find the place of each central pixel of the second window
                 % with minumum cost and substract the place of the initial
                 % central pixel of the first window. Save that value in the
                 % same place as for the initial central pixel of the first
                 % window
                 dispmap(i,j)=(k-j)*4;
              end
           end
           % this case is the same as before but it's for the pixel that are
           % less than 60 pixel to the end of the image
        else
           for  k=j:j+(450-j)-round(windowSize/2)
              Window_Try=Im_left(i-round(windowSize/2)+1:i+round(windowSize/2)-1,k-round(windowSize/2)+1:k+round(windowSize/2)-1);
              cost=sum(sum((Window_Try-Window).^2));
              if cost<maxcost
                 maxcost=cost;
                 dispmap(i,j)=(k-j)*4;
              end
           end
        end
     end
  end
  % show the image and cut the black pixels that we cant find the disparity
  % for them.
  dispmap=dispmap(round(windowSize/2):rows-round(windowSize/2),round(windowSize/2):columes-round(windowSize/2));
  subplot(2,2,1),imshow(mat2gray(dispmap));
  title('Disparity Map');
% read the ground truth depth map and cut edge rows and columes to be the
% same as the dispmap image
disp6=double(imread('disp6.png'));
disp6=disp6(round(windowSize/2):rows-round(windowSize/2),round(windowSize/2):columes-round(windowSize/2));
subplot(2,2,2),imshow(mat2gray(disp6));
title('Ground Truth Depth Map');
% compare the groun trouth with the calculated disparity map by their
% difference in pixel
diff=disp6-dispmap;
subplot(2,2,3),imshow(mat2gray(diff));
title('Difference by Subtracting Pixel Values');
% compare the groun trouth with the calculated disparity map by using
% matlabs 'diff' which creates a difference image from A and B.
subplot(2,2,4),imshowpair(disp6,dispmap,'diff');
title('Difference by Using Diff');
%calculating the corelation using the Pearson correlation to 2-D arrays
c = corr2(disp6,dispmap);
suptitle(['Correlation is ' num2str(c) ' for window size ' num2str(windowSize) ' pixels'] )

Best Answer

Optimizing code starts with finding the bottlenecks. It is not worth to optimize a part of the code, which takes 2% of the computing time only. If you increase its speed by a factor 2, the total program is just 1% faster.
The profiler helps to identify the lines, which needs the most time for your inputs. Without having your inputs (and readable code) we cannot do this for you. See:
doc profile
After you have found the bottleneck, you can post the relevant part of the code here. Add some inputs with typical sizes, e.g. created by rand(). Then the forum can help you to accelerate the program, most likely.
Related Question