Abstract: Aiming at the problems of being unsupervised and time-consuming of L1 norm optimization in the existing sparsity preserving projection,by integrating the sparse representation information with the manifold structure of the data,a novel algorithm for fast feature extraction,named sparsity preserving discriminative learning (SPDL),was proposed.SPDL first creates a concatenated dictionary by class-wise PCA decompositions and learns the sparse representation structure of each sample under the constructed dictionary using the least square method.Secondly,a local between-class separability function is defined to characterize the scatter of the samples in different sub-manifolds.Then SPDL integrates the learned sparse representation information with the local between-class relationship to construct a discriminant function.Finally,the proposed method is transformed into a problem of solving the generalized eigenvalue.Extensive experimental results on several public face databases demonstrate the effectiveness of the proposed approach.

Key words: Feature extraction,Sparse representation,Principal component analysis,Manifold learning,Face recognition