3D人体面部表情动画系统

3D Facial Animation System

本文讨论动画系统的一个子集的发展。中间目标是产生一个系统,它基于人体面部的视频输入,能实时产生人面的3D模型动画。首要的主题是模型操作、设计和配合、声频视频操作和同步、3D模型演示和表演、图像处理和特征提取。用连续的红绿蓝图像,在一个时间抓一帧画面,图像处理被计时。实验表明,用全面搜索定位面部特征大约需要70ms,用局部搜索定位特征大约5ms。考虑到模型显示,模型显示用OpenInventor成功完成。

This paper concerns the development of a subset of a system. The intermediate goal is to produce a system that can animate a 3D model of a human face in reahime, based on video input of a subject＇s face. The primary topic examined in this work is model manipulation, design and fitting, audio and video manipulation, synchronization, 3D model display and rendering, image processing and feature extraction. Using consecutive RGB images and grabbing one frame at a time, the image processing is timed. The test shows, that the facial features are located using the global search in approximately 70ms. Using the local search, the features were located in about 5ms. With respect to the model display, the model display is successfully implemented using Open Inventor.