基于张量脸的多姿态人脸识别算法  

Multi-view Face Recognition Based on Tensorfaces Algorithm

作　　者：姜珊[1,2] 王羽[1,2] 田春娜[1,2] 

机构地区：[1]中国石油大学(北京)机电学院,昌平102249 [2]西安电子科技大学

出　　处：《科学技术与工程》2009年第15期4381-4385,共5页Science Technology and Engineering

摘　　要：提出了一种新的多姿态人脸识别算法,在原有的张量脸算法(TensorFaces)基础上结合了流形学习方法和统计学聚类的方法,首先将训练图库中不同姿态的人脸图像通过保局映射投影(LPP)的姿态聚类特性投影到二维空间上,然后将待测图库中的未知姿态人脸图像投影到该二维空间并找到其最近邻的两个姿态,根据两个最近邻姿态库作为训练库修正张量脸识别算法的判别系数。实验结果表明,该算法的识别率优于原有的张量脸算法。An improved tensorfaces algorithm is proposed for multi-view face recognition which integrates multilinear analysis, manifold learning and statistical clustering in one framework. The training face images from different views are first mapped into a 2-D space by the Locality Preserving Projections （LPP） method where statistical clustering is used to capture the view variability. Then a test image of an unknown view can be projected into this 2- D space, and the two closet views can be identified. A modified tensor decomposition method is developed by incorporating two closest views as the new training database. The proposed method is evaluated on a large database of multi-view face images. Experimental results show that this method outperforms the original Tensorfaces method.

关 键 词：张量脸 流形学习 多姿态 人脸识别 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]