Synthetic Faces: Analysis and Applications

摘要

Facial animation has been a topic of intensive research for more than three decades. Still, designing realistic facial animations remains to be a challenging task. Several models and tools have been developed so far to automate the design of faces and facial animations synchronized with speech, emotions, and gestures. In this article, we take a brief overview of the existing parameterized facial animation systems. We then turn our attention to facial expression analysis, which we believe is the key to improving realism in animated faces. We report the results of our research regarding the analysis of the facial motion capture data. We use an optical tracking system that extracts the 3D positions of markers attached at specific feature point locations. We capture the movements of these face markers for a talking person. We then form a vector space representation by using the principal component analysis of this data. We call this space "expression and viseme space." As a result, we propose a new parameter space for sculpting facial expressions for synthetic faces. Such a representation not only offers insight into improving realism of animated faces, but also gives a new way of generating convincing speech animation and blending between several expressions. Expressive facial animation finds a variety of applications ranging from virtual environments to entertainment and games. With the advances in Internet technology, the development of online sales assistants, Web navigation aides and Web-based interactive tutors is promising than ever before. We overview the recent advances in the field of facial animation on the Web, with a detailed look at the requirements for Web-based facial animation systems and various applications.