3D Face Recognition Based on Local/Global Shape Description

The face recognition methods based on 2D images confront many challenges due to the variational conditions such as lighting, pose, facial expression and occlusion data. In this paper, we propose a new 3D face recognition approach using recent mesh parameterization techniques, Harmonic Mapping and ABF++. Different from existing 3D recognition methods, our approach represents the face surface in both local and global manners, which encodes the intrinsic attributes of surface in planar regions. Therefore, surface coarse registration and matching can be dealt with in low-dimensional space. Using a database of 61 3D face models with 327 partial scans, several experiments have been implemented for recognition under different pose, expression changes and occlusion data. The results show the feasibility of the proposed recognition scheme.