摘要
三维人脸识别是未来人脸识别的新方向,有望解决二维人脸识别的瓶颈问题;但三维人脸特征维数过高又制约了三维人脸识别的发展,特征降维意义重大。首先分析了传统降维算法的局限性和几种主要的流形学习算法,提出了将流形学习应用于三维人脸特征降维的思路,并构建了一个基于流形学习的三维人脸识别框架。
3D face recognition is the new tendency of face recognition,the bottleneck problem of 2D face recognition will be settled. But it is also be restricted because of the high feature dimension,feature dimension reduction is meaningful. This paper analyzed firstly the limitation of traditional method of reducing feature dimension and some main manifold algorithms. Presented a new mind of applying manifold to 3D facial feature dimension reduction,and designed a mainframe based on manifold for 3D face recognition.
出处
《计算机应用研究》
CSCD
北大核心
2010年第10期3982-3984,共3页
Application Research of Computers
基金
陕西省教育厅科研专项资助项目(09JK762)
