Automatic registration method for frontal 3D face

Three-dimensional face registration is a critical step in 3D face recognition. This paper presents an automatic registration method for frontal 3D face data. In our method, a unified coordinate system is created for each face according to the spatial relationships among its Symmetry Plane, nose tip point and nose bridge point. For evaluating the performance of our registration method, manually marked fiducial points are used for comparison. Extensive localization displacement and face verification experiments have been performed on the FRGC benchmark 3D face dataset to evaluate the accuracy and robustness of the proposed automatic registration method. Preliminary experiment results show that this approach can yield very accurate registration results with least human intervention.