Orthogonal Locality Sensitive Discriminant Analysis for Face Recognition
Yi Jin, Qiuqi Ruan · 2009
An innovative appearance-based method that called Orthogonal Locality Sensitive Discriminant Analysis is presented for face recognition in this paper. Our algorithm is based on the Locality Sensitive Discriminant Analysis (LSDA) algorithm, which aims at finding a projection by maximizing the margin between data points from different classes at each local area. However, a major disadvantage of LSDA is that LSDA is non- orthogonal, and this makes it difficult to estimate the intrinsic dimensionality and to reconstruct the face data. Non-orthogonality distorts the local geometrical structure of the data submanifold. In this paper, an Orthogonal LSDA algorithm is proposed to preserve the local geometrical structure by computing the mutually orthogonal basis functions iteratively. Since it produces orthogonal basis functions and can have more local structure preserving power, it is expected to have more discriminating power than LSDA. Experiments based on the ORL and Yale face database show the impressive performance of the proposed method. Results show that our new algorithm outperforms the other popular approaches reported in the literature and achieves a much higher recognition rate.