Entropy of subsections of an image

I have a grayscale image, which I am trying to take 3x3 sections of and find the entropy, and based on the value, make the centre pixel white or black.

The problem I have is that the code I am using is unacceptably slow - running a single, low res image takes several seconds. the code looks something like this:

for j = 2:m-1
   for i = 2:n-1
       E = entropy(im(j-1:j+1, i-1:i+1));
       if E < threshold
             ime(j, i) = 1;
       else
             ime(j, i) = 0;
       end
   end
end

where the grayscale image im is of size [m, n], as is the binary image ime. Any suggestions on how to improve it? Arrayfun may help, but i have no idea how to use it properly

Answer 1

[Iain]

Try

 E = entropyfilt(im,ones(3,3));
 ime = E < threshold;

[Simon]

That's exactly what I was looking for, thanks