How to insert image's features into Hidden Markov Model ?

Hi there,

Iam trying to compare between the two methods MLP/HMM in face recognition according to there accuracy,

I already know that MLP will do better, and HMM is good only in Speach recognition, but I have to try it anyway,

since iam new to matlab, i don't know functions and toolboxes ,

I've already donne the detection (using Viola-Jones), features extraction( std & mean) and training with the MLP (Multi-Layer Perception) parts,

but I only know some basics of HMM (Hidden Markov Model), i don't know how to implement my features (Data Base) into this system, I mean, what are the States & the Observations ? and how to use the features to train the system with HMM.

I'va made a capture of my DB, each row is a picture/face of a person that belongs to a class/label.

My question is how to put these features into the HMM system ?

Thanks in advance.

[ABDELKARIM MELIKI]

Anyone ?!