How to implement Crossing number technique for detection of retinal bifurcation point?

Hello I have to to extract feature( Bifurcation point, end points) from retinal vessels. From reading papers and some books I got to know that Crossing Number technique can be used for this. http://www.mnkjournals.com/ijlrst_files/Download/Vol%202%20Issue%202/22-6106-Sourav%20Saha.pdf http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6473297 http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7853731

Below is the image of the book i read for this purpose.

However I am not able to subtract adjacent pixel values in the boundery of 3 x 3 window. Below is really a simple code. I am able to extract 3 x 3 sub image from the main image. However I am not able to subtract adjacent pixel values in the bounder of 3 x 3 window.

%%Code
% a is input 
a= [ 1 0 1 1 1;
      1 1 1 1 1 ;
      1 1 1 1 1;
      1 1 1 1 1;
      1 1 1 1 1;];
%A for padding a at boundries
  A=padarray(a,[1,1]);
  for i =1: size(A,1)-2
  for j=1: size(A,2)-2

    win= A(i:i+2,j:j+2); % 3*3 Window from image
    for k=1: size(win,1)
        for m= 1:size(win,2)
       % CN
        end

    end

    end

    end