基于小波插值的人脸建模方法  

3D Face Modeling Based on Wavelet Interpolation

作　　者：许然[1] 汪亚明[1] 黄文清[1] 石信增[1] 

机构地区：[1]浙江理工大学,浙江杭州310018

出　　处：《计算机仿真》2011年第9期268-271,共4页Computer Simulation

基　　金：国家自然科学基金(50875245);淅江省自然科学基金(Y107558)

摘　　要：研究三维人脸建模问题,针对传统以二维图片为基础的三维人脸建模方法受到提取特征点过少的限制,不能很好的恢复三维人脸信息,导致三维人脸效果不佳的问题,提出了基于正交立体测量原理和二维高斯小波函数插值的三维人脸建模方法。方法运用正交测量的手段转化人脸二维坐标成三维人脸坐标,运用高斯小波函数进行空间人脸插值,增加三维人脸特征点,尽可能地恢复三维人脸结构信息。实验表明,方法最终增强三维人脸的数据模型的饱和度,达到了令人满意的效果。Research 3D face modeling. For two-dimensional face feature point extraction is limited, and can not restore three-dimensional face information accurately, a three-dimensional measurement principle is presented based on orthogonal and two-dimensional interpolation of three-dimensional Gaussian wavelet functions 3D Face Modeling method. This method uses orthogonal measurement to determine 2D coordinates to three-dimensional face coordi- nates. Gaussian wavelet interpolation method is used to restore three-dimensional structural information and increase three-dimensional facial feature points. The results show that the method can increase saturation of the data model and achieve satisfactory results.

关 键 词：特征点提取 立体测量法 高斯小波函数捅值 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]