Abstract
Animating lifelike human facial expression is a challenging task in computer graphics. There are two major unsolved problems in it. One is how to construct lifelike human facial models and another is how to animate these models. Traditionally, facial models have been build painstakingly by manual digitization and animated by ad hoc parametrically controlled facial mesh deformations. More recently, the 3D scanner has been used to digitize facial models and they have achieved more realistic animation through the dynamic simulation of facial tissues and muscles. However, the facial models which from 3D scanning still required user assistance to construct individual facial models suitable for animating facial expression. In this thesis, we present a construction and deformation method for head models with anatomical structure, suitable for real-time facial animation. Based on our facial modeling method, the skull is equipped with landmarks and skin thickness data of facial reconstruction and has the corresponding landmarks on skin, which allows us to deform the skull and skin in anthropometrically ways. We also have the muscle attachment points on skull and insertion points on skin, which allows us to automatically construct muscles. Finally, some feature points of MPEG4 on skin are used to deform our model to fit the individual facial model. On any deformed model, the underlying muscle and bone structure are adapted as well, such that the model is animatable using the same muscle contraction parameters. We also use these muscles to achieve the expression retargeting and editing. Moreover, we show that the advantages of our algorithm are not only to reuse expression data but also to generate new expression data if muscle model is applied.