Reserarch On The Techniques Of Face Modeling Based On Orthogonal Images

Building realistic 3D face models has always been a persistent challenging task in the field of computer graphics, image processing and computer vision. With the development of virtual reality, teleconference, movie making, computer animation, etc, people are paying more attention to the face modeling area. Aiming at the task of face modeling, this thesis focuses on the technique of face modeling based on orthogonal images. The research work can be organized as follows.Firstly, all kinds of techniques in the field of 3D face modeling and reflectance recovering was elaborated and compared, and then every different method fit for different application domain was introduced.Secondly, image-based personalized 3D facial modeling was discussed. An improved analysis-by-synthesis based personalized 3D facial modeling algorithm was given. Through measuring similarity between the texture synthesized and the original orthogonal images'texture, and then using this mis-registrations error to guide a local and adaptive subdivision step, we can refine the 3D facial mesh model to better preserve fine facial features, and make the output 3D facial model more realistic.Finally, a method for exploiting spatial coherence to estimate a non-parametric spatially-varying reflectance function-SBRDF from a sparse set of images even only one of known geometry was proposed. So we can render the face reflectance under new view point and new light environment or dynamic light source.