人脸特征约束点的三维表情合成  

New three-dimensional expression synthesis based on facial feature constraints points

作　　者：胡勇[1] 王国胤[1] 杨勇[1] 

机构地区：[1]重庆邮电大学计算机科学与技术研究所,重庆400065

出　　处：《计算机应用研究》2012年第2期754-756,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(61073146);重庆市杰出青年科学基金资助项目(2008BA2041);重庆市高等教育教学改革研究项目(09-1-004)

摘　　要：针对目前很多三维表情合成算法计算量大、方法比较复杂、真实感不强的特点,结合人脸分布特征,提出了一种基于Delaunay三角剖分的三维表情合成新算法。该方法通过对人脸特征点集的快速三角剖分避免了病态三角网格,有效提高了合成后人脸表情的真实度,降低了算法复杂度。大量实际人脸表情合成的实验结果表明,该算法通过少量的特征点生成的三维人脸表情更加真实,可有效快捷地合成各种真实的人脸表情。Aiming at a lot of three-dimensional facial expression of the synthesis method calculation was very large,complex and realistic not strong,this paper proposed the distribution characteristic of the face on the basis of a new Delaunay triangulations on three-dimensional expression synthesis method.The method was through a set of features to face rapid triangulations to avoid pathological triangular mesh.It could effectively improve the realism of facial expressions and reduce the complexity of the algorithm.A lot of actual face expression synthesis experiment results show that the proposed algorithm through a small amount of feature point generates three-dimensional facial expression more real,which can effectively fast synthesis of various real face expression.

关 键 词：三维人脸表情合成 DELAUNAY三角剖分 特征点 三维人脸模型 纹理映射 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]