Pose Variant Face Recognition using SPARSE l1- Regularized LS

As a recently proposed technique, sparse representation based classification (SRC) with l1-normminimization has been widely used for frontal face recognition (FFR). SRC first codes a testing sample as a sparse linear combination of all the training samples, and then classifies the testing sample by evaluating which class leads to the minimum representation error. But, it is really not the l1-norm sparsity that improves the FR accuracy.This paper analyzes the working mechanism of SRC, and indicates that it is the l1-regularized least squares formulation, with nonnegative constraints, that makes SRC powerful for face classification. Consequently, we propose a very simple yet much more efficient face classification scheme, for pose variant database namely ORL database with least square (LS). The extensive experiments clearly show that this scheme has very competitive classification results.