Fast and robust 3D face matching approach

We present a novel three-dimensional (3D) face matching approach in this paper. First, 3D facial scans are segmented and four feature points of each face are detected for rough alignment the by Absolute Orientation method. Then a modified Iterative Closest Point (ICP) algorithm is employed for range image registration. A Simulated Annealing (SA) based approach with the Surface Interpenetration Measure (SIM), as similarity measure, is used for final matching. Our experimental results on the Face Recognition Grand Challenge (FRGC) v2 database show that the proposed method could achieve 99% rank one recognition performance at 0.001 False Acceptance Rate (FAR) on all neutral expression with noisy data.