Training on Statistical Feature Models of Action Units for 3D Facial Expression Recognition

Most of the existing studies on 3D Facial Expression Recognition (FER) are messaged-basedapproaches, which only detect the already known six universal expressions. In this paper, wedescribe the group of global and local features used to comprehensively characterize facialactivities. These features are further used to train Statistical Feature Models (SFMs) associatedwith each Action Unit (AU). The occurrence probability of a specific AU on an input textured3D face model is then computed. The results demonstrate that the evidence of AUs is ofimportance for applying AU space to evaluate expressions.