Abstract: To solve the problem of low complexity of low-rank matrix recovery algorithm and occlusion of training set samples, this paper proposes an occlusion face recognition method based on low-rank matrix recovery and Gabor feature robust representation and classification. Firstly, the fast low-rank matrix recovery algorithm is used to accurately and quickly obtain the error image corresponding to the training sample image. Then, Gabor transform is performed on the "clean" face image and the occlusion error image respectively to obtain the Gabor feature vector. Then, In this paper, an occlusion dictionary compression algorithm based on Gabor feature is proposed. The compressed Gabor occlusion dictionary can be calculated and used to form a Gabor compression dictionary with Gabor feature vectors of training samples. Finally, the test samples are cooperatively represented by the compression dictionary. The final recognition result. The experimental results on Extended Yale B and AR database show that the proposed method not only has strong robustness to occlusion face recognition, but also greatly reduces the computational complexity of occlusion face image coding and reduces the running time of the algorithm.