3D Face Recognition based on Local Conformal Parameterization and Iso-Geodesic Stripes Analysis

Abstract

3D face recognition is an important topic in the field of pattern recognition and computer graphic. We propose a novel approach for 3D face recognition using local conformal parameterization and iso-geodesic stripes. In our framework, the 3D facial surface is considered as a Riemannian 2-manifold. The surface is mapped into the 2D circle parameter domain using local conformal parameterization. In the parameter domain, the geometric features are extracted from the iso-geodesic stripes. Combining the relative position measure, Chain 2D Weighted Walkthroughs (C2DWW), the 3D face matching results can be obtained. The geometric features from iso-geodesic stripes in parameter domain are robust in terms of head poses, facial expressions, and some occlusions. In the experiments, our method achieves a high recognition accuracy of 3D facial data from the Texas3D and Bosphorus3D face database.