An improvement on face recognition rate using local tetra patterns with support vector machine under varying illumination conditions

Varying illumination condition make face recognition a challenging issue. In this paper, we propose a novel approach for solving this problem by performing three steps: (1) illumination normalization that normalizes the images using pre-processing chain of Gamma Correction, Difference of Gaussian and Contrast Equalization; (2) local tetra patterns (LTrPs) for texture based face representation; and (3) Support Vector Machine (SVM) for classification. The experimental results shows that face recognition rate of proposed system is improved from 97.9%, 97% and 99% to 99.34% as compared with LDP using Histogram Intersection, LBP using Chi-square, and LTP using Euclidean Distance respectively on Extended Yale-B database.