Virual Two-Man Act: Lip And Expression Synthesis For 3D Face Animation

3D facial reconstruction and facial animation synthesis of painting is a hot research topic in the field of the current computer graphics, computer vision and image processing , in video conferencing, video production, communications have a wide range of applications, has great application value and commercial value. This paper presents a human face lip and facial animation synthesis system, detailing the structure of the system involved in face detection, facial modeling, feature selection, feature extraction, such as facial animation technology. This article will face in the 3D reconstruction based on the synthesis of facial animation, including facial expression synthesis and synthesis of lip animation.This synthesis is based on the animated face model CANDIDE-3. Construction for the accurate model, based on Haar features and the AdaBoost algorithm for face and eyes regional monitoring methods, and then extracted through the eyes of gray projection characteristics, for CANDIDE-3 model reconstruction parameters. Side face image through the depth of access to model data to construct 3D face model.To complete the three-dimensional face of the lip synthesis, extraction of lip feature points required the movement of data, this work completed in the corpus CAVDatabaseⅡ. This article is about selecting the lip of the mouth, upper and lower lip boundary point feature extraction motion data. In this work, this paper outlines based on level set motion feature extraction method, achieve a good results.This animation using MPEG-4 standard, in three-dimensional model and the lip on the basis of the data, respectively, realistic 3D and two-dimensional cartoon head face was lip synthesis. Using MPEG-4 model, a model of a three-tier structure animation framework, to achieve the definition of AUV and 6 of 16 kinds of basic expressions, and ultimately the superposition of expressions and lip animation. Finally, the whole system flow diagram, the system interface and realistic 3D face and two-dimensional animated cartoon picture of the final synthesis results.