On the basis of research for sparse representation and Gabor wavelet, a new method for face recognition combined Gabor with representation is proposed in this paper. Gabor wavelet transformation is used for face image from the training image set to obtain facial features, the over-complete dictionary is built by the Gabor features from all training set, and the sparse facial feature is obtained by sparse representation algorithm. Finally, the classifier based on fusion is designed for face recognition. The experimental results show that the improved method can extract facial features and structure information more effectively, and it also can improve face recognition rate greatly.