Sparse Representation Based Face Recognition Under Varying Illumination

Illumination with different lighting levels or angles is an important problem for classification of an individual in face recognition. To overcome this issue, generally classification algorithms are applied after pre-processing of the images to get rid of the low contrast regions to increase the accuracy of recognition. In this work, we use Steerable Gaussian Filter at the pre-processing step for all training and testing samples. In the classification step, we use the recently proposed “Classification via Sparse Reconstruction Vector (CSRV)“ algorithm. The performance of our approach is compared with that of the “Principal Component Analysis (PCA)” algorithm in terms of recognition rates (RR). Experiment results show that the CSRV algorithm has a better performance than that of the PCA algorithm with higher RR even for poorly illuminated images taken from Yale Database B.