3D face registration on simulated annealing algorithm

Face registration is a critical step for 3D face recognition. This paper studies face registration based on a simulated annealing algorithm using the depth information of a 3D face. In our method, the annealing algorithm is adopted to select the appropriate fitness-function - MLESAC and SIM. A simple algorithm which involves three steps is proposed to perform the critical step of 3D face recognition. Simulation results show that the simulated annealing algorithm can escape from local optimal solutions, and converge to the global optimal solution quickly. Furthermore, MLESAC and SIM helps to control the registration process effectively, and improve the accuracy of recognition.