Array Correlation - compute array difference with its neighbor elements

Hi,

Do you know the fastest way to compute array mean difference with its neighbor elements in a N x N matrix? and is there a possibility to perform for second neighbors, third and so on.

Suppose you have

 a=[1 4 5 3 6 ; 8 10 0 9 11 ; 2 5 20 15 9 ; 8 12 4 6 0 ; 7 9 18 2 5]

so I want a code to give this result for mean difference of first neighbors:

d(1,1)=[(1-4)+(1-8)]/2, d(1,2)=[(4-1)+(4-5)+(4-10)]/3, d(2,2)=[(10-8)+(10-0)+(10-4)+(10-5)]/4 and etc. For second neighbors: dd(3,3)=[(20-2)+(20-9)+(20-5)+(20-18)/4].

thank you

Answer 1

[Walter Roberson]

mask = [0 1 0; 1 0 1; 0 1 0];
Ncount = conv2(ones(size(a)), mask, 'same');
d = double(a) - conv2(double(a), mask, 'same')./Ncount;

You can leave out the double() if you are sure that a will not be integer data type (images are usually integer data type.)

For second difference, use an appropriate larger mask such as

[0 0 1 0 0; 0 0 0 0 0; 1 0 0 0 1; 0 0 0 0 0; 0 0 1 0 0]

[ichi]

Thank you for the answer, I knew the conv2(a,mask,'same'), gives you the sum of the neighbors of each element, but this method is so good.

[Walter Roberson]

If the question is resolved, please Accept the answer.