Implementing equation doe 3x3 matrix

I have a image below ,please tell how to implement the equation for a 3x3 matrix

http://imgur.com/6NO9M

please assist

Answer 1

[Matt J]

qVMF=median(q(:));

[Walter Roberson]

Will the median necessarily be the one that minimizes the sums ? Wouldn't the pixel that is closest to the mean be the one that minimizes the sums ?

[Matt J]

Yes, the median is the known solution to this problem, which you can also verify by comparing it with the output of the code in your Answer.

The mean minimizes the sum of squared differences.

It would get slightly complicated if there were an even number of pixels, because then the median wouldn't be attained at any of the q(i). Here, however, there are an odd number of pixels.

Answer 2

[Walter Roberson]

[minsum, minidx] = min( sum(abs(bsxfun(@minus, q(:), q(:).'))) );
qVFM = q(minidx);