With the emergence of online payment, smart home, and automated attendance, efficient and convenient identity authentication can greatly improve our quality of life. Among them, face recognition technology has received more and more attention from the society, and more and more face recognition algorithms have been proposed. Because LBP (local binary patterns) is simple in calculation and strong in feature classification, it is widely used by many researchers in face recognition. In this paper, the traditional LBP algorithm and existing improvements are investigated, and the characteristics of the LBP algorithm are verified by experiments. According to some shortcomings of the current LBP algorithm, several improved schemes are proposed. The recognition rate and speed of the improved LBP algorithm are tested on the ORL (Olivetti Research Laboratory) database to verify the performance improvement.