Face recognition using kernel entropy component analysis.pdf

Neurocomputing 74 (2011) 1053–1057Contents lists available at ScienceDirectNeurocomputing0925-23 doi:10.1 n Corr E-m sharmil natalia.djournal homepage: /locate/neucomLettersFace recognition using kernel entropy component analysisB.H. Shekar a, M. Sharmila Kumari b,n, Leonid M. Mestetskiy c, Natalia F. Dyshkant c a Department of Computer Science, Mangalore University, Mangalore, Karnataka, India b Department of Computer Science & Engineering, PA College of Engineering, Mangalore, Karnataka, India c Department of Computational Mathematics and Cybernetics, Moscow State University, Moscow, Russiaa r t i c l e i n f o Article history: Received 7 June 2010 Accepted 17 October 2010Communicated by L.C. Jain the best principal component vectors that are subsequently used for pattern projection to a lower- dimensional space. Extensive experimentation on Yale and UMIST face database has been conducted to Available online 17 December 2010 Keywords: Principal component analysis Entropy component analysis Eigenface Face recognition12/$ - see front matter & 2010 Elsevier B.V. A 016/j.neucom.2010.10.012 esponding author. ail addresses: bhshekar@ (B.H. Shek abp@ (M. Sharmila Kumari), mestlm yshkant@ (N.F. Dyshkant).a b s t r a c t In this letter, we have reported a new face recognition algorithm based on Renyi entropy component analysis. In the proposedmodel, kernel-basedmethodology is integratedwith entropy analysis to choose reveal the performance of the entropy based principal component analysis method and comparative analysis is made with the kernel principal component analysis method to signify the importance of selection of principal component vectors based on entropy information rather based only onmagnitude of eigenvalues. & 2010 Elsevier B.V. All rights reserved.1. Introduction In the current decade and the last decade,wehavewitnessed the tremendous growth of research in the field of biometrics, in particular on face recognition because of its wide acceptability in several