Matlab: Image Processing through Expectation Maximization

I need to implement Expectation Maximization through the use of Matlab for image segmentation. The basic idea is below from previous work done. A framework of codes and a possible set of EM codes can be provided but modifications need to be done. Not only would I need a working set of codes but I'll need guidance to understand both EM and the codes fully. Idea: A sequence of images(eg. 20 frames) input into the framework codes which make use of EM to produce background models.(eg. 3 background models each with a mean and variance) A current(test frame) image is being compared with the background models. 3 D-matrices are produced using D=I-mu/sigma (something like this) and the min is taken : D=min(D(:,:,1), D(:,:,2), D(:,:,3)); Thresholding is done next (i.e. map=<5) Blobs are formed for connected component analysis. Small regions are removed for better performance.