Abstract
Recently, face recognition has become one of the most widely research areas in the biometric identification domain. Its popularity is due to that it is less intrusive than other biometric systems thus making it highly be accepted by people. Although many papers reported face recognition methods, such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). These methods have achieved high recognition rates for several public face image databases. However, the noise have deep influence on recognition rates. A robust face recognition based on compressive sensing, called a Sparse Representation-based Classification (SRC) method was proposed. In this thesis, we compare PCA, LDA, and SRC on three publically available face databases: the JAFFE, the ORL, and the FEI databases. Experimental results show that SRC reaches the higher recognition accuracy when the number of training images is small.