一种参数化的表情映射方法  

An Algorithm for Parameterized Expression Mapping

作　　者：吴渊[1] 郑文庭[1] 

机构地区：[1]浙江大学国家重点实验室,浙江杭州310027

出　　处：《计算机应用研究》2004年第10期117-119,121,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(60103017)

摘　　要：人脸表情细节的合成是生成高度真实感人脸动画的重要环节,传统的表情映射技术仅考虑了人脸特征点的位置变化,无法处理表情细节;最新的研究考虑了表情纹理细节的合成,但仅能根据已有的表情样本一对一地获取纹理细节,应用范围很有限。有效地结合小波金字塔分解和表情比例图像两种技术,提出了一种进行表情合成的新算法。该算法既可以映射表情纹理细节,又可以生成表情动画,且表情夸张程度可参数化地控制,算法简单有效。The synthesis of human expression texture,is an important factor in realistic animation of human face.Traditional expression synthesis technologies are only concerned about the geometrical transformation of facial features.However,they do not deal with expression details.Recent researches try to involve synthesis of expression texture,but most of them can only generate expression details from given expression samples using one- to- one method,leading to the limited application areas.A new algorithm is raised by effectively combining wavelet pyramid decomposition and expression ratio image.This algorithm can not only map the expression texture,but also generate expression animation.The expression exaggeration factor is also parameterized so that the user can control the expression freely.The algorithm proves to be effective.

关 键 词：表情映射 小波分解 人脸动画 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]