Parameterized Facial Expression Synthesis for Videoconferencing Applications

Abstract

In this paper we propose a method of creating intermediate facial expressions based on primary ones [1]. To achieve this goal we utilize both Facial Definition Parameters (FDPs) and Facial Animation Parameters (FAPs). We introduce a way for modeling the primary expressions using FAPs and we describe a rule-based technique for the synthesis of intermediate ones. Furthermore, a relation between FAPs and the activation parameter proposed in some classic psychological studies is established. In this way we try to get advantage of the extended work that have been done from psychologists and which covers much more expressions than the archetypal ones the computer society concentrated on. The overall scheme leads to a parameterized approach of synthesizing facial and can be used for the creation of MPEG-4 compatible synthetic video sequences