Research Of Facial Expression Simulation Based On Hybrid Model

Facial expression plays a vital role in people's intercommunion, and serves as an important way for people's non-verbal intercommunion. With the rapid development of information technology and the unceasing enhancement of human dependence on computer, facial expression simulation, as a way to promote the exchange ability between people, obtains more and more attentions from researchers. Since the first 3D facial expression animation establishes by Parke in 1972, the face modeling and facial expression simulation technology has progressed by leaps and bounds in the past 30 years, and has made a great deal of research results. It widely applies at different fields, including the entertainment, the communication, the medicine, the education and so on. This paper, on the basis of previous research, through establishing specific face model, using related knowledge on human anatomy and biomechanics, proposed a hybrid model used for face lattice deformation, and implemented the realistic simulation of facial expression.Based on comparing the methods of representation and construction of 3D face model, according to the requirements of 3D face model in real-time facial expression simulation, face model exported from Poser 7.0 were adopted and optimized to serve as the general face model. then for the given front and lateral photos of faces, radical basis function was used to carry on facial adaptation. The mosaic weighted factor in facial texture synthesizing was improved, and was combined with image mosaic method based on Gauss-Laplace pyramid, then the seamless multi-resolution panoramic facial texture was achieved. In order to implement the texture mapping of face model, Delaunay triangulation algorithm based on incremental Insertion algorithm was used to establish the mapping relations between face model and texture.On comprehensive analysis and comparison of the previous research results, a hybrid model for facial expression simulation was proposed using related knowledge on human anatomy and biomechanics. This model mainly includes three sub-models: elastic skin model, muscle model and skeleton model. In the elastic skin model, the nonlinear mass-spring model was used to simulation the facial elasticity behavior, and edge repulsion spring was joined to avoid the unrealistic phenomenon of face model. Muscle serves as one of the most important factors for generating facial expression, the force conditions of face lattice vertices are different according to the type of the muscle, so the force conditions were analyzed and force formulas were given. In order to simulate jaw's movement, head skeleton model was joined to. It can not only simulate jaw's movement by simple calculation, but also play a constraint role during the facial deformation. So, it can generate more realistic facial expressions by using the hybrid model.Finally, the face lattice deformation equation based on hybrid model was given, and a facial expression simulation system was designed and implemented. The system holds great expansibility. Experiments indicate that it can not only implement the realistic simulation of different facial expressions, but also reach a speed of more than 50 frames per second which goes far beyond the requests of real-time animation.