Multi-Scale Primal Feature Based Facial Expression Modeling and Identification

In this paper, we present our newly developed face expression modeling system for expression analysis and identification. Given a face image at a front view, a realistic facial model is created using our extended topographic analysis and model instantiation approach. Our facial expression modeling system consists of two major components: (1) facial feature representation using the coarse-to-fine multiscale topographic primitive features and (2) an adaptive generic model individualization process based on the primal facial surface feature context. The algorithms have been tested using both static images and facial expression sequences. The usefulness of the generated expression models is validated by our 3D facial expression analysis algorithm. The accuracy of the generated expression model is evaluated by the comparison between the generated models and the range models obtained by a 3D digitizer.

• Type: Conference
• Year: 2006
• Where: FGR
• Authors: Lijun Yin, Xiaozhou Wei
• Keywords: Biometrics, Expression Modeling, Facial Expression, FGR 2006, Generated Expression Model