Face Biometric Identity Authentication System

With the popularity of payment by smartphone, mobile financial transactions security becomes serious. This paper takes smartphone as application background, combineing with several algorithms to propose a face identity authentication system. Firstly, samples are preprocessed. Then, Haar and AdaBoost algorithms are adopted for face detection, features of face area are extracted by LBP operator. Finally, Euclidean distance is used for feature matching to achieve authentication. The test results show that preprocessing suppresses the illumination influence effectively, face detection accuracy reach to 97.2% and 96.6% respectively, system authentication accuracy can be as high as 98.86%, FRR and FAR are kept at a low value of 1.14% and 0.89%, the average authentication time takes 271~327ms. In summary, the face identify authentication system proposed in this paper has better robustness and higher authentication accuracy while maintaining good timeliness.