A Comparative Study on Vector-based and Matrix-based Linear Discriminant Analysis

Recently a kind of matrix-based discriminant feature extraction approach called 2DLDA have been drawn much attention by researchers. 2DLDA can avoid the singularity problem and has low computational costs and has been experimentally reported that 2DLDA outperforms traditional LDA. In this paper, we compare 2DLDA with LDA in view of the discriminant power and find that 2DLDA as a kind of special LDA has no stronger discriminant power than LDA. So, why 2DLDA outperforms LDA in some cases? Through theoretical analysis, we find it is mainly because of the difference of stability under nonsingular linear transformation and linear operation power between 2DLDA and LDA. In experimental parts, the results of experiments give enough proof on our claims and show in some cases the performance of 2DLDA will be possible superior to that of LDA and in other cases the performance of LDA will be possible superior to that of 2DLDA.

[1]  Pong C. Yuen,et al.  Regularized discriminant analysis and its application to face recognition , 2003, Pattern Recognit..

[2]  Palaiahnakote Shivakumara,et al.  (2D)2LDA: An efficient approach for face recognition , 2006, Pattern Recognit..

[3]  M. Omair Ahmad,et al.  Two-dimensional FLD for face recognition , 2005, Pattern Recognit..

[4]  David J. Kriegman,et al.  Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection , 1996, ECCV.

[5]  Jieping Ye,et al.  Two-Dimensional Linear Discriminant Analysis , 2004, NIPS.

[6]  Xiaoou Tang,et al.  Dual-space linear discriminant analysis for face recognition , 2004, CVPR 2004.

[7]  E. K. Teoh,et al.  Generalized 2D principal component analysis , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[8]  Ja-Chen Lin,et al.  A new LDA-based face recognition system which can solve the small sample size problem , 1998, Pattern Recognit..

[9]  Bin Luo,et al.  2D-LPP: A two-dimensional extension of locality preserving projections , 2007, Neurocomputing.

[10]  Pengfei Shi,et al.  A note on two-dimensional linear discriminant analysis , 2008, Pattern Recognit. Lett..

[11]  Hanqing Lu,et al.  Solving the small sample size problem of LDA , 2002, Object recognition supported by user interaction for service robots.

[12]  Xingzhi Zhan.  Matrix Inequalities , 2002 .

[13]  Jieping Ye,et al.  Feature Reduction via Generalized Uncorrelated Linear Discriminant Analysis , 2006, IEEE Transactions on Knowledge and Data Engineering.

[14]  Konstantinos N. Plataniotis,et al.  Regularization studies of linear discriminant analysis in small sample size scenarios with application to face recognition , 2005, Pattern Recognit. Lett..

[15]  Ming Li,et al.  2D-LDA: A statistical linear discriminant analysis for image matrix , 2005, Pattern Recognit. Lett..

[16]  Alejandro F. Frangi,et al.  Two-dimensional PCA: a new approach to appearance-based face representation and recognition , 2004 .

[17]  Jing Peng,et al.  Discriminant Analysis: A Least Squares Approximation View , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Workshops.

[18]  Jian Yang,et al.  From image vector to matrix: a straightforward image projection technique - IMPCA vs. PCA , 2002, Pattern Recognit..

[19]  Jian-Huang Lai,et al.  1D-LDA vs. 2D-LDA: When is vector-based linear discriminant analysis better than matrix-based? , 2008, Pattern Recognit..

[20]  R. Chellappa,et al.  Subspace Linear Discriminant Analysis for Face Recognition , 1999 .

[21]  A. Hmamed,et al.  A matrix inequality , 1989 .

[22]  Hakan Cevikalp,et al.  Discriminative common vectors for face recognition , 2005, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[23]  Xin Pan,et al.  Palmprint recognition with improved two-dimensional locality preserving projections , 2008, Image Vis. Comput..