3D Facial Animation On Multilayer Structure Muscle Model

Human face is the window of communication in real world, and virtual face is of the same importance in virtual world. Therefore, the construction and animation of realistic human facial models is always attractive to researchers in computer graphics, and as an identifiable area of the latter in the past 30 years, facial modeling and animation has been devoted to significant efforts with many dozens of research papers published. Its applications include such diverse fields as character animation for films and advertising, computer games, virtual reality, teleconferencing, user-interface surgical facial planning, and criminal discrimination. Yet no perfectly realistic facial model and animation has ever been generated by computer, it remains one of the most difficult problem in computer graphics.Human face synthesis is divided into two main branches: facial modeling and facial animation. The main challenge lies in facial modeling and facial animation and much research work focuses on it .Since the anatomic structure of human faces is complicated, and people are familiar and sensitive with appearance of human faces, it is difficult to synthesize realistic human facial animation, including both geometry and texture. Otherwise, how to represent and organize existing facial expression data effectively in order to reuse it on individual faces is also an interesting topic. At last, making the system more automatic and decreasing manual work of users is also a goal of such technique. The main work of this thesis focuses on follow two points: 1. In view of face modeling we present an automatic method of the personalization of face model. First, Cyber ware laser scanner scans the face of a specific person to acquire its geometry data. Then, Then, we generate a face model which with 863 points. And use developed Hoppe optimization guide line to make the basic face model match to the scanned face model. This arithmetic generates a realistic face model by using little manual work. Last, texture mapping is used to map the textureimage on the face model. A customization face model is generated and can be rendered.2. In view of mimicking the face expressions and motions, this paper presents a new physically-based 3D facial model based on anatomical knowledge which provides high fidelity for facial expression animation while optimizing the computation. Our facial model has a multilayer biomechanical structure, incorporating .a physically-based approximation to facial skin tissue, a set of anatomically-motivated facial muscle actuators, and underlying skull structure. Through these models natural looking animations have been able to generate.The work and achievement in this thesis is just fundamental and pilot study, while it can be exploited in many applications such as natural and intelligent human-machine interfaces, virtual collaboration work, virtual learning and teaching, and so on.