Improving Robustness and Time Efficiency with Weight Function in Occlusion Face Recognition

The need for occlusion face recognition in daily life, professional fields like criminal investigation and the recent epidemic is increasing, so it is still a hot issue in the current research and many researchers are pursuing the technical excellence for the face recognition of random occlusion. In the previous research, RRC model is proposed for the excellent performance in the occlusion face recognition and the IR
³
C algorithm is adopted to solve this model. Among them, the weight function in this algorithm plays a big role, which will directly affect the recognition rate and running time of the experiment. In this paper, three weight functions: Inverse trigonometric function, Hyperbolic tangent function and Log function are proposed for the purpose of improving the robustness and time efficiency of IR
³
C algorithm. Applying these functions on the comparable types and levels of random occlusion and then comparing them with the previous research, the experimental results indicate that the inverse trigonometric function is superior to other functions with better recognition rates and shorter running time.