Constructing A Realistic Face Model of An Individual for Expression Animation
Yu Zhang, Edmond C. Prakash, Eric Sung · 2002
This paper presents a method for creating photorealistic textured 3D face models of specific people for dynamic facial expression animation. The modeling approach reconstructs an accurate geometrical face model based on the individual face measurements containing both shape and texture information that is acquired from a laser range scanner. By using a semi-automatic registration and merging technique, a 3D dense face mesh is recovered fro m the partial range data obtained from arbitrary multiple different views. A model editing and adaptive meshing scheme is then used to refine the surface model. Having recovered the facial geometry, we add realism by mapping the model with high-resolution texture images. The resultant synthetic face has been shown to be visually similar to the true face. Based on the geometrically accurate surface model, a physically-based face model with a hierarchical structure of the skin, muscles and skull is developed from anatomical perspective. The dynamic displacement of nodes in the skin lattice under the influence of internal muscular forces is calculated by a numerical integration method. Using our technique, we have been able to generate highly realistic face models and flexible expressions.