Realistic Texture Mapping Of Facial Mesh Model

With the continuous development and improvement of computer graphics technology, the virtual reality technology is widely used in film and television animation, advertising design, games and entertainment, biological feature recognition and other aspects increasingly. Creating realistic specific3D facial model has been a challenging task among the numerous research areas. Due to the complexity of human facial features, how to ensure that the model and the facial feature matching, and the model to achieve realistic texture mapping is a key problem, also is the focus of this paper. The research contents include:1. Illumination model Building.Adding lighting effects to the scene based on the realistic image rendering process and OpenGL illumination model which improved the reality of the3D model showing in the scene. By adjusting the respective parameters proportion of the ambient light, diffuse light, specular light and emissive color in the scene, then calculate the results to form the final illumination effect.2. Facial feature extraction. Implement face detection based on Skin-color model.At first,calculate skin similarity for the two-value setting,then determine the general range of face according to the results, again correct distortion due to the algorithm limitations by the artificial interface, finally extract face contour and feature information.3.3D model adjustment.Propose a model adjustment algorithm based on the morph proportion. Firstly, divide the3D model into blocks from top to bottom, and then calculate each block-width through the abscissa of the maximum and minimum points, according to the width and height of the model to calculate the proportion. Similarly, calculate the proportion of every line about the image which extracted the face contour. Lastly, interpolate three-dimensional coordinates of3D model to match the face contour on the basis of these proportions.4. Texture mapping of3D model.This paper propose a constrained texture mapping method,which can maintain the alignment of the corresponding features in the texture mapping process by the predefined constrained points.Firstly,calculate the displacement difference between two key points.then change the corresponding relations between the original texture coordinates and3D model coordinates according to the displacement.Finally,with feature points accurately matched, texture mapping to make the model more realistic.Furthermore, we designed and developed a texture mapping system based on the VC++6.0and graphics library OpenGL. The system is able to generate3D model of target face by entering a single face image for face detection, feature extraction, and model adjustment. It has carried on the constrained texture mapping and light regulation, which make the results more realistic.