新的局部线性嵌入下的人脸识别方法
|
摘要
针对LLE算法无法对后续采集的测试样本单独进行降维处理和未能利用样本点分类信息的两点不足之处,提出了一种有监督的增量式局部线性嵌入算法(SILLE),并采取小波变换对图像进行预处理。通过对ORL数据库实验证明,SILLE算法与LLE算法相比大大降低了处理新增样本点的计算时间,并且提高了识别精度。
In view of the disadvantages that LLE algorithm is unable to follow-up the test samples which are collected sepa-rately and to use dimensionality reduction,LLE algorithm does not make use of the classification of information sample point as well,a method is proposed to recognize the face using Supervised Incremental Locally Linear Embedding algorithm(SILLE),and combined with wavelet transformation to pretreat the images.Experimental results based on ORL database show that SILLE algorithm compared with LLE algorithm greatly reduces calculating time spent in handling additional sam-ples and improves recognition accuracy.
In view of the disadvantages that LLE algorithm is unable to follow-up the test samples which are collected sepa-rately and to use dimensionality reduction,LLE algorithm does not make use of the classification of information sample point as well,a method is proposed to recognize the face using Supervised Incremental Locally Linear Embedding algorithm(SILLE),and combined with wavelet transformation to pretreat the images.Experimental results based on ORL database show that SILLE algorithm compared with LLE algorithm greatly reduces calculating time spent in handling additional sam-ples and improves recognition accuracy.
引文
[1]Belhumer P N,Hespanha J P,Kriegman D J.Eigenfaces vs fish-erfaces:recognition using class specific linear projection[J].IEEETrans on Pattern Anal Machine Intell,1997,19(7):711-720.
[2]Hyvrinen A,Oja E.Independent component analysis:algorithmsand applications[J].Neural Networks,2000,13(4/5):411-430.
[3]Lu J W,Plantaniotis K N,Venetsanopoulos A N.Face recogni-tion using LDA based algorithms[J].IEEE Transactions on Net-works,2003,14(1):195-200.
[4]Feng G Y,Hu D W,Zhou Z T.A direct locality preserving pro-jections algorithm for image recognition[J].Neural Processing Let-ters,2008,27(3):247-255.
[5]Roweis S T,Saul L K.Nonlinear dimensionality reduction by lo-cally linear embedding[J].Science,2000,290(5500):2268-2269.
[6]冈萨雷斯.数字图像处理[M].阮秋奇,阮宇智,译.2版.北京:电子工业出版社,2006.
[7]杨绍华,林盘,潘晨.利用小波变换提高基于KPCA方法的人脸识别性能[J].山东大学学报:理学版,2007,42(9):96-100.
[8]钟广军.基于提升方法的图像压缩技术的研究[D].长沙:国防科学技术大学,2000.
[9]Dick D R,Olga K,Oleg O,et al.Supervised locally linear embed-ding[C]//Artificial Neural Networks and Neural Information Pro-cessing,ICANN/ICONIP 2003.Germany:Springer,2003:333-341.
[10]Kouropteva O,Okun O,Pietikainen M.Incremental locally lin-ear embedding[J].Pattern Recognition,2005,38(10):1764-1767.
[11]王振华.基于非线性维数约减的模式识别[D].天津:天津大学,2004.
[12]何怀玉.LLE算法在地震属性参数降维中的应用[D].成都:成都理工大学,2006.
[2]Hyvrinen A,Oja E.Independent component analysis:algorithmsand applications[J].Neural Networks,2000,13(4/5):411-430.
[3]Lu J W,Plantaniotis K N,Venetsanopoulos A N.Face recogni-tion using LDA based algorithms[J].IEEE Transactions on Net-works,2003,14(1):195-200.
[4]Feng G Y,Hu D W,Zhou Z T.A direct locality preserving pro-jections algorithm for image recognition[J].Neural Processing Let-ters,2008,27(3):247-255.
[5]Roweis S T,Saul L K.Nonlinear dimensionality reduction by lo-cally linear embedding[J].Science,2000,290(5500):2268-2269.
[6]冈萨雷斯.数字图像处理[M].阮秋奇,阮宇智,译.2版.北京:电子工业出版社,2006.
[7]杨绍华,林盘,潘晨.利用小波变换提高基于KPCA方法的人脸识别性能[J].山东大学学报:理学版,2007,42(9):96-100.
[8]钟广军.基于提升方法的图像压缩技术的研究[D].长沙:国防科学技术大学,2000.
[9]Dick D R,Olga K,Oleg O,et al.Supervised locally linear embed-ding[C]//Artificial Neural Networks and Neural Information Pro-cessing,ICANN/ICONIP 2003.Germany:Springer,2003:333-341.
[10]Kouropteva O,Okun O,Pietikainen M.Incremental locally lin-ear embedding[J].Pattern Recognition,2005,38(10):1764-1767.
[11]王振华.基于非线性维数约减的模式识别[D].天津:天津大学,2004.
[12]何怀玉.LLE算法在地震属性参数降维中的应用[D].成都:成都理工大学,2006.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心