南京大学学报(自然科学版) ›› 2012, Vol. 48 ›› Issue (1): 115122.
• • 上一篇
宣士斌**
Xuan Shi-Bin
摘要: 特征抽取是模式识别中的一项重要工作,其中特征抽取的子空间为一法一直受到研究者的关注,特别是近此年来研究人员提出各种加权子空间为一法,但这此权重都是人为设定.为此,提出一种权重
自动学习算法,该算法以缩小学习样本到其所属类原型的距离同时增大学习样本到其它类原型的距离为学习目标,在两个为一向上调整权值,保证了算法收敛.同时,钊一对主成分分析及线性判别分析的变形最
大边缘准则,重新定义了它们对应的带权协为一差矩阵和带权散布矩阵,该定义充分表达了权重的本质含义.在3个公开人脸数据库上的实验室结果显示提出的算法有更好的识别率与更高的稳定性.
[1]Matthew T,Alex P. Eigcnfaces for recogni tion.Journal of Cognitive Neuroscience,1991,3(1):71~86 [2]Chin T J,David S. Incremental Kcrncl principal component analysis. IEEE Transactions on Im- arc Processing, 2007,16(6):1662一1674. [3]Zhu M L. Subcalss dicriminant analysis, IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(8):1274~1286 [4]Chen I. F, Liao H Y M, Ko M T,et al. A new LDA-based face recognition system which can solve the small sample size problem. Pattern Recognition, 2000,33(1):1713一1726. [5]Friedman J H. Regularized discriminant analy sis. Journal of the American Statistical Associa tion, 1989,84(405):165一175. [6]Hastie T,Tibshirani R. Penalized discriminant analysis.The Annals of Statistics 1995,23(1): 73一102. [7]Zhuang X S, Dai D Q, inverse Fisher discrimi- nant criteria for small sample size problem and its application to face recognition. Pattern Rec ognition, 2005,38(11);2192一2194. [8]Yang J,Zhang D, Yang J Y. Face recognition using Laplacian faces, IEEE Transactions on Pattern Analysis and Machine intelligence,2007,29(4)650一664. [9]He X F, Cai D, Niyogi P.Tensor subspace analysis. Advances in Neural information Pro ceeding Systems,Vancouver, Canada, 2005,18:5~7 [10]Li H,Jiang T,K,Efficien and robust feature extraction by maximum margin criteri-on, IEEE Transactions on Neural Networks 2006,17(1):1157一1165. [11]Zhao D, Lin Z C, Xiao R,et al. Lincar Lapla- cian discrimination for feature extraction. Pro- cecdings of the international Conference on Computer Vision and Pattern Recognition, Min- neapolis, 2007,1一7. [12]Baudat G. Anouar F. Generalized discriminant analysis using a kernel approach. Neural Com putation, 2000,12(10);2385一2404. [13]Yang W K,Wang J G, Ren M W, et al. Fea ture extraction using fuzzy inverse FDA. Neuro computing, 2009,72(13):3381一3390. [14]Yang W K,Wang J U, Rcn M W, et al. Fea- ture extraction based on Laplacian bidirectional maximum margin criterion. Pattern Recogni- tion, 2009,42(11):2327一2331. [15]Zhang Z, Yc N, Du H,et al. Sulrpattci-based discriminative semi-supervised nonlinear dimen- sionality reduction. Journal of Nanjing Univer- say(Natural Sciences), 2009,45(5):693一603.(张召,业宁,杜辉等.子模式判别型半监督非线性维数减少算法.南京大学学报(自然科学),2009,45(5): 693一603). [16]Georgia Tech Facc Database,ftp;//ftp, ee, gat eeh,edu/pub/users/hayes/facedb/. [17]Chen L, Man H,Nefian A V. Face recognition hased on multi-class mapping of Fisher scores. Pattern Recognition, 2005,38(6):799一811. [18]ORL facedatabase.http://www.cam-orl.co.uk [19]Martinez A M,Benavente R. T he A R face da tabase. Computer Video Centre Technical Re port 24.1998 |
No related articles found! |
|