A Realistic Human Face Mouth Animation Research

Realistic 3D facial expression mouth animation research has been an important research direction in voice visual modeling. Especially with the development of Internet technology and data processing technology, how to narrow the distance between man and virtual human to humanize the behavior and thought of virtual human, and enhance friendly interaction between virtual human and man is an important development direction in computer science.On the basis of a lot of research productions on realistic facial modeling, how to realize coarticulation between voice and facial lip animation has been an urgent problem. The study of realistic facial lip animation and visual voice modeling technique include the following questions. Firstly, the realistic face model generally only considers modeling human facial region and the modeling process rarely makes detailed modeling on mouth and eyes. As a result, the established model is not sufficient to express complex motion effects during mouth speaking. Secondly, on the study of the synthesis of mouth animation, establishing muscle model to compound the mouth animation is a common way. Since during muscle modeling process, it is necessary to establish the appropriate muscle movement control parameters for each muscle. The muscle around the mouth will cause the problem of excessive control parameters. The mouth animation is the result of the muscles synergistic effect. It is not easy to distinguish weight control for each muscle, and access muscle control parameters. Basing on mouth motion decompositions to compound mouth animation is a relatively simple way, but researchers rarely consider the intrinsic link between child movements. There are some problems existed, such as excessively careful divisions and overlapping motion effects in sub-movements. Thirdly, during the process of coarticulation phenomena modeling, the interaction relationship usually be considered only in phonemes inside syllables, with little regard for the interaction between syllables, especially for the interaction relationship between vowels.For the above problems, a refining facial mesh model based on the standard Candide-3 model is established in this thesis. We synthesize a real face model through single photograph. Then, by analyzing the internal relation on characteristics between the muscle movements and sub-movements, a mouth motion controlling model based on the motion decomposition is established, and the motion change control of basic speaking mouth shape is realized. Finally, we start from the constituent structure of Chinese phonetic alphabets to analysis the phoneme of coarticulation, to establish a visual consonant effect weight function inside syllables and to create a visual weight function on syllables between the interior vowel syllables. From the established visual weight function, the interaction levels of pronunciation mouth between the phonemes can be described. Finally, the synchronization of a real facial lip animation and speech is achieved. The experimental analysis indicates that the realistic facial mouth animation proposed by this paper is more close to the variation features of real mouth pronunciation.