3D Face Modeling Based On Person's Photos

Since Parke generated face image with the face parameter model in 1972, 3D face modeling has been an important offset of computer graphics. It can be used in more and more fields: virtual reality, film production, man-machine interaction interface, video telephone, video conference, computer aided edu- cation, medicine research and computer games, et al. Although researchers have achieved many productions, modeling reality face of a given person is all the same a difficult task because of the complexity of face surface.This paper focuses on the technology of 3D face modeling based on person's photos. First, get the 3D file of a general face from the modeling software Poser 6.0, and transform the data to the format adapted to the later study. Then photos of the modeling object are selected from the face photo library, extract the face feature points from these photos and save them. This prepares for the next steps. Following, the general face model is personalized using the space interpolation technology of RBF and Kriging. To improve the result of Kriging, the 3D model is smoothed by interpolating once again in a local area near every feature point, a better natural result is achieved. Next, for getting the face texture information, this paper tries to use the image Pyramid and Wavelet to stitch the person's photos. During stitching, a kind of betterment is presented, which stitches three photos of the same person instead of two photos. It avoids bigger distortion of texture got by two photos. A 360°face texture map independent of view is made using the improved method. Finally, Map the texture to the gridding model, an ideal 3D face model characterized is displayed.In the phase of personalized geometry shape of the general face model, this paper uses Matlab 6.5 to implement a face modeling system. Using its functionof data exportation, the final 3D face with texture is painted in Visual C++ 6.0 using OpenGL. Taking advantage of Matlab 6.5 and Visual C++ 6.0, a favorable study direction and valuable experience for face modeling is proposed.