Face Recognition Based on Sparse Representation Classifier with Gabor-Edge Components Histogram

We describe a new method for recognizing humans by their face, which is robust to the variations of facial imaging conditions, with high accuracy. The human face recognition system consists of three components: i) a new face descriptor based on edge component histogram and its variance between pixels, ii) Gab or-edge components histogram for facial image representation, combining the Gab or wavelet and the proposed edge components histogram, iii) a sparse representation classifier for the face recognition. The effective and robust face recognition with high accuracy is achieved by the Gab or-edge components histogram and the sparse representation classifier. In experiments, higher face recognition performances, which are 99.45% on ETRI database and 99.41% on XM2VTS database, have been achieved.