Combining contrast information and local binary patterns for gender classification

Abstract

Recent developments in face analysis showed that local binary patterns (LBP) provide excellent results in representing faces. LBP is by definition a purely gray-scale invariant texture operator, codifying only the facial patterns while ignoring the magnitude of gray level differences (i.e. contrast). However, pattern information is independent of the gray scale, whereas contrast is not. On the other hand, contrast is not affected by rotation, but patterns are, by default. So, these two measures can supplement each other. This paper addresses how well facial images can be described by means of both contrast information and local binary patterns. We investigate a new facial representation which combines both measures and extensively evaluate the proposed representation on the gender classification problem, showing interesting results. Furthermore, we compare our results against those of using Haar-like features and AdaBoost learning, demonstrating improvements with a significant margin. © 2011 Springer-Verlag.