Based On The Mpeg-4 3d Face Animation Design And Realization

The trend of computer multi-media development requests more and more directand perspicuous communication between human and computer. Realistic 3D humanfacial modeling and its animation which is widely used in movie making, pc games,video conference, and man-machine interface has always been the active research areain computer graphics and man-machine conversation.In this thesis I will focus on two aspects: 3D face modeling and face animation.In view of face modeling, some former methods of face modeling has been studiedand compared. On the base of that, an idea of modularization modeling has beenproposed just as follows: first of all, an initialized and neuter face model will be built;then grid nodes will be selected as the control points of face modeling according to thefacial organ function zones, and improved DFFD arithmetic will be used to generatecustom-built face model; finally, software prototype will generate large number ofdifferent face models that could be selected by experimenters. This modeling methodwill decrease the calculation task of DFFD greatly, control the deformation processexactly to generate the living face model and be convenient for software modeling.In view of mimic the face expressions and motions, a new method of faceanimation based on MPEG-4 is introduced. Firstly, we design a multi-layer FAP drivingprecept and expand the FADT; then we divide the facial expressions and motions intotwo parts: rigid body's movement and skin texture movement. This method can carry onthe lifelike simulation to the real face movement. Finally, we make good use of FAPframe with mask code to calculate the mid-frame in order to decrease the data amountand increase the reality of facial expressions. By using of COM transmission model, itis unnecessary for experimenters to consider the detailed low level. In a word, real faceexpressions will be realized by adopting such method.Lastly, we develop a 3D face animation system compliant to MPEG-4, which couldsimulate the face expressions and movement under the FDP and FAP streams drive.