Conformal Parameterization and Curvature Analysis for 3D Facial Recognition

摘要

This work proposes a new algorithm for 3D face recognition. The algorithm uses 3D shape data without color or texture information and exploits local curvature information which is a measure with high discriminant capability and robust to deformations such as rotation and scaling. In order to reduce high dimensionality of typical face surfaces our approach uses a conformal parameterization, preserving angles of original faces and simplifies the correspondence problem. Experimental results are presented and discussed using CASIA and Gavab databases.