Fisher locality preserving projections for face recognition

In this paper, a novel dimensionality reduction method termed Fisher Locality Preserving Projections (FLPP) is proposed by introducing the maximum scatter difference criterion (MSDC) to the objective function of Locality Preserving Projections (LPP). FLPP not only inherits the advantages of LPP which attempts to preserve the local structure, but also makes full use of class information and orthogonal subspace. After being embedded into a low-dimensional subspace, the samples of the same class maintain their intrinsic neighbor relations, whereas the samples of the different classes are far from each other. In addition, the small sample size problem (SSS) is avoided. As a result, the most discriminative feature is extracted. Experiment results on the ORL and FERET face databases demonstrate the effectiveness of the proposed FLPP method.