摘要
文章介绍了如何把基于径向基函数多步离散数据插值的方法用在人脸变形技术上,该方法是在对脸部定义的特征点进行分层的基础上,和已知特征点位移的条件下,利用一种基于迭代的插值方法求出特征点周围的网格点位移,从而进行人脸变形。这种方法既保证了脸部局部区域变形的精确性和平滑性,又减少了运算的复杂度。在人脸动画方面该算法也得到了广泛的应用。
This paper proposes that how to use the compactly supported radial basis functions for deformation algorithm in facial calibration.Under the condition of known original coordinates and displacement of facial definition parameters(FDPs ),this method based on iteration and hierarchizing FDPs to compute the displacement of mesh points around the FDPs.So it's convenient to implementation in facial calibration.This method not only assures the accuracy and smooth-ness in local area of face,but also decreases the algorithm complexity of general deformation techniques.In addition,this method also is widely applied to facial animation and so on.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第24期80-82,120,共4页
Computer Engineering and Applications
基金
安徽省自然科学基金资助(编号:01042203)
中国科学技术大学青年基金资助
