Robust Face Recognition

Abstract

Sparse representation (SRC) based Face classification has been developed extensively in last decade due to its superior performance compared to the other existing methods. Sparse representation of a query sample as a linear combination of all the training samples, and evaluation in terms of classification leads to the minimal representation error. Most of the SRC scheme uses l1 norm constraint in the classification, while all training samples, to collaboratively represent a query sample has been ignored. This thesis deals with the development of composite sparse and collaborative representation based classifier (CRC). The proposed method is robust against various illumination change, pose change of the test image. The sparse based collaborative representation provides excellent classification as well as very low computation, due to composite scheme. This thesis employees feature extraction process using LBP and extended LBP method. Extensive simulations are done for the assessment of the proposed methods. From the simulation results it has been found that the proposed methods yields better performance as compare with that of state of-the-art methods.