A Support Vector Machines Approach for Efficient Facial Expression Recognition

Abstract

Current scenario in computer vision demands an efficient and robust technique for facial expression recognition. There is also a need for a generalized technique that can even be used for content based image retrieval and analysis. This paper introduces a novel methodology of facial expression recognition using Support Vector Machines. An efficient model is trained and developed using the necessary features extracted by employing 2D Gabor filters. Practically, six different methods for handling the feature vectors are discussed and extensively analyzed in this paper. The developed model is tested and cross validated and the detailed results are presented. It is observed that the proposed method offers a consistent and good accuracy (83.3%) for all the six basic expressions considered. In addition, the implementation complexity is reduced by minimizing the number of support vectors, unlike the traditional counterparts. The proposed method shall definitely turn out to be an effective alternative for the existing methods.