Extraction and alignment of facial regions for 3D facial image-based recognition

Face recognition is the process of identifying a person using his image as biometric data. 3D image-based face recognition is expected to overcome many problems that are faced in traditional 2D face recognition, such as the lack of explicit shape information, pose and lighting variations. However, as a relatively new technology, it does face a number of challenges among them are: 1) How to automatically extract the facial region from the data captured by a 3D facial scanner, which may contain hair, neck and shoulder and possibly other background data and 2) How to precisely and efficiently align two instances of 3D faces. This paper proposes new innovative techniques that address some aspects of these challenges.