what type of normalization is this?????

X=im2col(X,[8 8],'sliding');

X=(X-repmat(mean(X),[size(X,1) 1]));

X=X ./ repmat(sqrt(sum(X.^2)),[size(X,1) 1]);

Answer 1

[Walter Roberson]

It operates on an 8 x 8 sliding window. Within each window, the mean is subtracted and then the data is divided by the square root of the sum of squares of the values. The result is going to be a column vector of length 64 with mean 0 and whose sum of squares is 1, for each window (unless the 8 x 8 block contained all 0 in which case you get back NaN). But remember that for some reason it is doing a sliding window.

I would say that this is meant to prepare each sliding window for feature detection. It is not intended that this be reconstructed into an image array.