Face Recognition Using Hidden Markov Modelling

This paper presents a hybrid framework of feature extraction and hidden Markov modeling (HMM) for two-dimensional Face recognition. Importantly, we explore a new discriminative training criterion to assure model compactness and discriminability. This criterion is derived from the hypothesis test theory via maximizing the confidence of accepting the hypothesis that observations are from target HMM states rather than competing HMM states. We have two stages first stage is feature extraction using DCT and second stage for training HMM is used. From the experiments on the FERET database and GTFD, we find that the proposed method obtains robust segmentation in the presence of different facial expressions, orientations, and so forth. In comparison with the maximum likelihood and minimum classification error HMMs, the proposed HMM achieves higher recognition accuracies with lower feature dimensions.