Abstract: As one of the most attractive biometric techniques,face recognition is still a challenging task.This is mainly owing to the varying lighting,facial expression,pose,and environment.In this sense,a face image is just an observation and it should not be considered as the absolutely accurate representation of the face.However,even in a real world face recognition system,it is difficult to obtain enough samples.The great success of sparse representation in image reconstruction triggers the research on sparse representation based pattern classification.Inspired by this,a sparse representation based classification method using category elimination and greedy search strategy was proposed for face recognition.First,we reduced the uncertainty of the face representation by synthesizing the virtual training samples.We applied an error-constrained orthogonal matching pursuit algorithm to exploit an optimal representation result of training samples from the classes by eliminating the category and the specific training samples.The final remaining training samples are used to produce a best representation of the test sample and to classify it.Then,we selected useful training samples that are similar to the test sample from the set of all the original and synthesized virtual training samples.Finally,we devised a representation approach based on the selected useful training samples to perform face recognition.Experimental results on five widely used face databases demonstrate that our proposed approach can not only obtain higher face recognition accuracy,but also has a lower computational complexity than the other state-of-the-art approaches.

Key words: Sparse representation,Greedy algorithm,Face recognition,Classifications