Personalized Face Modeling System Based On A Single Frontal Face Photos

In view of 3D virtual human face have wide application prospect in the digital entertainment, video communications and identity authentication and many other industries. Realistic face modeling is currently a hot and difficulty research in computer graphics. Many graphics experts are working to build a fully realistic three-dimensional model which simulation face. The paper gives a comprehensive survey and analysis of the existing facial modeling techniques and methods. On the basis of comparing existing technologies and methods of facial modeling, In-depth study the mature modeling technology, we put forward own facial modeling method and designs and realizes individuality facial modeling system.We try to model a realistic 3D face model from a single photograph of front view. First, we use camera get a neutral face photos, and then use the active shape model (ASM) detected the68 feature point's position of man's face. These feature point can calibrate the man's face feature information, for example, the nose, mouth information, eye, etc; then select a generic face models, mark the feature point in the model according to the feature point on the photo position, thus established the corresponding of 2d and 3d. Next, we design and realize the personalized face modeling system. Due to user-centric, we create the photo processing function modules. The module's function is Format conversion and scaling operations for user input, make the photo to mapping requirements. We embedded the ASM algorithm in the system; the input photos will be automatically detected. And then we create the manually modification the function module which can adjust the position of the detected feature points. Experiments show that the adjustment for the feature point which be detected on the bad light and posture photographs that can get a more realistic face model. After adjustment, the 68 feature point of the face photograph converted to 3D coordinates, as the X and Y values of the facial model. The depth values Z is to adopt a common z value of the experience points corresponding to the model. Transform of the 3D feature points RBF interpolation function through the general model fit to the individual face model. The 3D feature points are obtained after transformation which fit the generic model into individual face model through the RBF interpolation function. At the same time, we can obtain the texture coordinates through the RBF interpolation function. Last, we extract the texture from picture, and texture mapping using OpenGL technology for the model, and finally generate a realistic face model personalization. This paper provides a simple and quick method for face modeling and the design and implementation of face modeling system is simple to use, with good interaction, for any user to operate the mouse can be used.