Application And Research On SVM And HMM Hybrid Model

Face Recognition is a study region with high academic and applicative value. It is important in human communication that to get a man's information on his face. So that, Face Recognition is one of the most potential identify method. With its endlessness application, it became the hot point in pattern recognition.Support vector machine is a powerful machine-learning method, and obtains better performance than traditional methods in the applications of multi-dimension nonlinear pattern classification. Hidden Markov Model is a statistic model and usually be used to describe signal features. Its elementary theory is proposed by Balm and Welch in 20th 60s, and be frequently used on speech recognition. In this paper, a hybrid SVM/HMM model for face recognition is constructed by means of embedding SVM into the framework of HMM. Additional, several issues that arise as a result of the hybrid framework have been addressed including estimation of posterior probabilities and the use of segment-level data.To grapple with the blemish of classic HMM, we use backdated HMM (BHMM) instead of classic HMM in this paper. We compose a hybrid Model HBS (Hybrid BHMM and SVM) with BHMM and SVM. Experiments show that the hybrid model could get better performance.