Image-driven 3d Face Animation Research

Computer Gfraphics, Images processing, and Computer visions are three kinds of computer science related to human being visions. In the above three areas, virtual facial modeling and expression animation are always the most difficult and important parts. In the dissertation, I have completed the system of facial animation that image drove. In addition, I describe the details of the technology of building up this system, such as detecting face, detection and tracking facial feature, face modeling and so on, as well as the analyses of the details of different technologies.First of all, for the face detection, I make use of the face detection based on the evolving Agent, so as to confirm the face's area, and the method based on facial knowledge and edge extracting to localizes eyes and mouse, and then, I adopt the method based on the deformable template and snake, to realize facial feature tracking. In addition, I discuss the label problem in the part of face modeling characters. In this thesis, I use the definition of MPEG-4 for reference, and adjust it to the need of face characters. Finally, I achieve utilizing the changes of face feature to drive the facial animation. I obtain the controlling information from the facial feature detection of the sequences, and adopt the information to drive the character points on modeling, carry out the free-form deformation, and thus realize the animation. Basically, this system has achieved the anticipated effect, realizing the facial animation.