MOTIVATING CLASS-SPECIFIC NONLINEAR PROJECTIONS FOR FACE VERIFICATION

In this paper we motivate the use of class-specific non-linear subspace methods for face verification. The problem of face verification is considered as a two-class problem (gen-uine versus impostor class). The typical Fisher’s Linear Dis-criminant Analysis (FLDA) gives only one or two projections in a two-class problem. This is a very strict limitation to the search of discriminant dimensions. As for the FLDA for N class problems (N> 2) the transformation is not person spe-cific. In order to remedy these limitations of FLDA, exploit the individuality of human faces and take into consideration the fact that the distribution of facial images, under differ-ent viewpoints, illumination variations and facial expression is highly complex and non-linear, novel kernel discriminant algorithms are used. The new method is tested in the face verification problem using various datasets where it is verified that it outperforms other commonly used kernel approaches Index Terms — Face verification, two-class problems, ker-nel techniques, Fisher’s linear discriminant analysis