A new face recognition algorithm using LBP and NFS

Real-time identity verification and recognition of celebrity´s web facial image used in special occasion such as access control and computer security, etc, is a tough issue in face recognition. And it has attracted more and more people´s attention recently. This paper proposes a novel effective and real-time face recognition method for solving this problem. Owing to enhance the accuracy ratio, coarse-to-fine structure technique used for gradually improving the recognition ability by improved iteration method which further analyzes the intermediate result of candidates of the coarse stage, is adopted in this paper. The simple and effective Local Binary Patterns (LBP) technique and Nearest Feature Space (NFS) classifier that has the good generalization ability are combined to construct a new way for satisfying the real-time and robustness decision performance. The face image database derived from web is used for our experiment. The experimental results show that the proposed method performs better than traditional approaches in terms of real-time efficiency (0.7s per image) and accuracy (72.5% Average Recognition Rate).