Realistic 3d Facial Animation Based On Mpeg-4 Implementation

Animation of realistic human facial model is an important research field in computer graphics in recent 30 years. The application fields include virtual presenter, virtual meeting, computer assisted teaching, medical research, film making, and computer game etc. The research on this field has great theory and application value.There are two challenges for the animation of realistic human facial models. One is to construct a perfect face model; the other is to generate realistic face expressions.How to simulate the motions of human faces on 3D facial models in real-time to generate realistic facial expressions is a challenging problem. Based on the theory of facial animation in MPEG-4, a new method of realizing 3-D facial animation is proposed in this thesis. This method is applicable for all topological face models. With the drive of FAP frames, the proposed method can realize realistic facial animation. The generated data is of small size, compatible with all MPEG-4 system, also suitable for network transportation. The system realized with this method has great advantage for user's operation.Constructing a realistic face model is another challenge for realistic face animation. In this thesis, a new method of constructing 3-D face model is proposed, which could randomly generate lots of different 3-D face models. With this method, user can save his favorite face model without much manual work. Users are also allowed to do some small change to the mode. Applying the deformation algorithm, this method can relieve user from boring manual work. As the foundation of deformation method, proper design of general model and control points, are presented in this thesis.