AAM Facial Feature Points Positioning And Application In Expression Animation

Active Apparent Model (AAM for short) is one of the main and effective methods that used for facial feature points positioning, and it has been studied by many scholars and also been used in many facial systems.This paper purposed to research AAM facial feature points positioning technology and its application in facial expression animation. It proposed a performance driven facial animation system based on the AAM. The system can automatically obtain the facial expression motion information from the face images or video, and then drive the MPEG-4 face model for synthesis of facial expression animation. System mainly divided into three modules:AAM facial feature points positioning module, FAP conversion module, and MPEG-4 model animation module. This paper focuses on the front two modules.AAM facial feature points positioning module is the master key for its orientation result directly influence the synthetic effect of the facial expression animation. This paper first introduced the AAM theory, and then made some improvement based on the shortcomings of the traditional AAM algorithm. Compared with the traditional algorithm, improved AAM has obvious advantage in fitting accuracy and arithmetic speed algorithm and has more practical.FAP conversion module is necessary, and is the interface module for AAM technology and MPEG-4 standard. It can not only improve the data compression ratio of generated animation, but also can enlarge the application scope.This paper's innovation is the proposing of the AAM based performance driven facial animation system, and boldly combining AAM technology and the FAP parameters in MPEG-4 for driving models to make facial animation. Although at present some scholars already combined AAM technology and MPEG-4 standards for facial animation, but AAM are mostly used to establish the corresponding specific MPEG-4 model, the motion data used to drive the MPEG-4 model are get by other methods。...