The automatic face recognition system has become one of the most important functions in the field of computer vision. The systems of recognition have attracted great attention in terms of different applications in human life. Facial recognition based on automatic methods to determine the identity of individuals by different changes of the face, as indicated in the database [ORL]. In this article, we compare two techniques for measuring the classification rate. The first one is based on the extraction of key points by both methods (PCA-SIFT and SURF), then the association by the RANSAC algorithm, finally we determine face authentication based on distance measurement (Euclidean and Manhattan). The second one is based on the extraction of local features by three local binary patches (TPLBP) and then classification by support vector machine (SVM). Simulation results show that the second technique gives a good result reaching 98% and increases the recognition rate by 1.468%