Abstract

Facial animation finds applications in very diverse areas of our lives. It is used in entertainment industry (cinema, video games) as well as in more "serious" industries: virtual humans can be used for medical purpose (in prediction of plastic surgery or in speech therapy) or in remote learning and teaching. Although accurately modeling of the human face is still an important challenge in facial animation, it seems that in the last few years research about human-computer interaction dominates in facial animation. Among research on man-machine interaction, the important area of research is how to built a system with a "human face" which would be able to replace a real person in a conversation. All current existing systems for automatic generation of facial expressions are rule-based. That means that a set of rules describe dependences between the text, emotions and facial expressions. This set of rules is based on the work of many psychologists, who tried to describe universal links between facial expressions and the verbal content of a message. The whole idea of the system for generating facial animations is based on a "facial expression script language". In 1970 psychologists P. Ekman and F.W. Friesen developed an universal Facial Action Coding System (FACS) based on Action Units (Aus), where each facial expression can be described as a combination of Aus. Their work became de facto a standard in analysis of facial expressions not only in psychology but also in computer facial-related topics, that's why it has been decided to base facial movements on this system.