Multi pose face recognition using double stages classifications: SMLDA and fusion of scale invariant features

This paper presents alternative technique for multi-pose face recognition using double stages classifications: shifting mean LDA (SMLDA) and fusion of scale invariant features (FSIF) based face descriptor. The first stage is employed to find the best class candidates that are similar to the query image the second stage (i.e FSIF) is employed to find the best matched class corresponding to the query image. The aims of this method are to solve the large face variability due to pose variations, to decrease the computational time of FSIF-based face recognition, and to avoid using the 3D scanner for estimating any pose variations of a face image without decreasing the recognition performance. The experimental results show that proposed method can overcome large face variability due to face pose variations, need short the computational time, and give better recognition rate than those of the previous method. In addition, the proposed method also provides better recognition rate than that of 3D based methods without requiring 3D scanner.