Surface normal angle for 3D face recognition

摘要

A common 3D face recognition method is to align two faces together and perform surface matching. Although to achieve a good alignment is a time consuming task, this step is vital to obtain a good recognition rate using surface matching. Therefore, in this paper, a 3D face recognition method that does not require vigorous alignment is proposed. This method consists of using the surface normal of the face and the identity is determined by calculating the angle differences between the surface normal of 2 faces. Surface normal is chosen since the value would be the same regardless of the face pose. The face with the smallest angle difference is determined to be the identity of the unknown face. Simulation results show that this proposed method is able to produce good recognition rates and does not require the usage of fine alignment.