Face detection using HMM -SVM method

Abstract

This paper proposes a method for face detection and recognition using Modified Hidden Markov Model (HMM) and Support Vector Machine (SVM). It is a two layer architecture system that identifies all image regions which contain face or non-face. At the first stage, the Kernel HMM classifies input pattern into three classes: a face class, undecided class or non-face class. In the final stage, SVM detects the face class or non-face class if any sub-image falsely judged as undecided class. This system alleviates the problem of false positive rate. The experimental result shows that the proposed approach outperforms some of the existing face detection methods and we have compared various face detection method. © 2012 Springer-Verlag GmbH.