基于人眼多线谱的MPEG-4参数计算  

MPEG-4 Parameter Computing with Multi-curve Spectrum of Eyes

作　　者：曾丹[1] 程义民[1] 陈志翔[1] 钱振兴[1] 

机构地区：[1]中国科学技术大学电子科学与技术系,合肥230027

出　　处：《数据采集与处理》2009年第2期159-164,共6页Journal of Data Acquisition and Processing

基　　金：中国科学技术大学研究生创新基金(KD2006037)资助项目

摘　　要：提出了由人眼多线谱计算MPEG-4相关人脸动画参数(FAP)的方法。人眼多线谱利用一组随时间变化的归一化肌肉收缩量表示人眼运动及表情,通过对其采样可获得视频各帧对应的归一化肌肉收量。根据三维肌肉控制人眼(3D-MCE)模型与MPEG-4标准模型在特征点、运动表示参数、坐标及尺度上的对应关系,可由归一化肌肉收缩量集计算相应FAP,以驱动任意基于MPEG-4标准的人眼模型。实验证明该方法所用数据量及计算量较小,所得MPEG-4参数可以驱动得到较逼真的人眼运动及表情。A method for computing face animation parameter （FAP） of MPEG-4 is presented with multi-curve spectrum of eyes. Multi-curve spectrum describes eye movements and expressions with a group of time-varying curves of normalized muscle contractions. According to sampling the multi-curve spectrum, the normalized muscle contractions for each frame are obtained. Based on the correspondence relations among the 3D muscle controlled eye（3D-MCE） model and MPEG-4 at feature points, animation parameters of eye movements and expressions, and coordinates and scales, FAP for the any MPEG-4 eye model is computed with the corresponding sets of normalized muscle contractions. Experimental results show that the method has the low calculating complexity and the small data quantity. Simulation results can obtain eye movements and some expressions.

关 键 词：虚拟人脸 肌肉收缩量 MPEG-4 多线谱 三维肌肉控制人眼模型 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]