人脸识别的一种新的特征提取方法  

Face recognition based on a new feature extraction method

作　　者：祝磊[1] 朱善安[1] 

机构地区：[1]浙江大学电气工程学院,浙江杭州310027

出　　处：《光电工程》2007年第6期122-125,共4页Opto-Electronic Engineering

摘　　要：针对人脸识别中判别特征的提取问题,本文提出了一种新的人脸识别算法—扩展保局投影(ELPP)。普通保局投影(LPP)在构建权图时侧重保持样本的局部结构,属于无监督学习算法。扩展保局投影在保局投影的基础上进行扩展,通过引入可调因子,在保持人脸图像局部流形结构的同时考虑样本的类别信息,从而充分提取样本的判别特征。本文采用最小近邻分类器估算识别率。在Yale人脸库以及AT&T人脸库的测试结果表明,在姿态、光照、表情、训练样本数目变化的情况下,ELPP都具有较好的识别率。In view of the discriminant feature extraction problem in face recognition, a new face image feature extraction and recognition method-Extended Locality Preserving Projections （ELPP） is proposed in this paper. When constructing graphs, LPP emphasizes face sample manifold local structure belongs to unsupervised learning algorithms. By using turning parameter, ELPP combines both the face manifold local structure information and class label information, and extracts the discriminant feature of face for recognition. The proposed method was tested and evaluated in the Yale face database and AT＆T face database. Nearest Neighborhood （NN） algorithm was used to construct classifiers. The experimental results show that ELPP has good performance when pose, lighting condition, face expression and train sample number change.

关 键 词：保局投影 人脸识别 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]