Matlab: Window based Otsu thresholding

This function divides the gray level image into a predefined set of subimages (This can be controlled). Calculates the otsu threshold on each subimage and threshold it accordingly. The function returns a BW image that is thresholded according to window based otsu thresholding.

function IMBw = LocalOtsuThresh( IMGray, NW )

[M N] = size( IMGray );
IMBw = zeros(M, N);

if nargin < 2
    NW = 4;
end

for m = 1:NW
    for n = 1:NW
        IMTmp = uint8(IMGray( 1+(m-1)*floor(M/NW):m*floor(M/NW),...   
            1+(n-1)*floor(N/NW):n*floor(N/NW) ));
        IMTmpBw = im2bw( IMTmp, graythresh( IMTmp ) );
        
        IMBw( 1+(m-1)*floor(M/NW):m*floor(M/NW),...
            1+(n-1)*floor(N/NW):n*floor(N/NW) ) = IMTmpBw;
    end
end

NW defines the number of subimages. Default is 4. So in default case the gray level image will be divided into 16 subimages.