The recognition and simulation of conversational facial expressions at various levels of photorealism

点击翻译按钮获取中文摘要

Communication is simultaneously one of the most important and one of the most complex tasks that humans undertake. Facial motions play a central role in the process of communication, and can be used to help modify what is said, to control the flow of a dialog, or to convey complex intentions without saying a word. In this project, we aim to use a synergistic combination of psychophysical and computer graphics methodologies in order to develop a systematic description of the facial motions that are required to produce realistic, natural looking, conversational facial animations. In one line of research, we will employ advanced facial animation techniques to determine which areas of the face need to move (and which are sufficient) for an expression to be recognized. The information that is used to recognize an expression, or to determine its believability must be dependent on the information that is actually present. The perception of expressions may, then, be dependent on the level of realism. The project will examine the perception of expressions at different levels of abstraction, from real video footage, through highly realistic animations down to highly non-photorealistic animations.