3-D face verification with 3-D face scanner

Abstract

A 3-D to 3-D face verification technique is described in the paper. A practical 3-D face scanner that captures both texture and range images is used for registration and verification. The posture of an input 3-D face image is normalized for matching it to the registered 3-D face image. The criterion of stableness of face region is defined for the prior verification. The stable regions are considered to be useful for the matching and the verification. In the experiment, the stable face regions are analyzed and a superior verification rates are obtained.