Please use this identifier to cite or link to this item:
Title: Spatially smooth subspace face recognition using LOG and DOG penalties
Authors: Zuo, W
Liu, L
Wang, K
Zhang, D 
Keywords: Derivative of Gaussian
Face recognition
Laplacian of Gaussian
Subspace analysis
Issue Date: 2009
Publisher: Springer
Source: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics), 2009, v. 5553 LNCS, no. PART 3, p. 439-448 How to cite?
Journal: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics) 
Abstract: Subspace face recognition methods have been widely investigated in the last few decades. Since the pixels of an image are spatially correlated and facial images are generally considered to be spatially smoothing, several spatially smooth subspace methods have been proposed for face recognition. In this paper, we first survey the progress and problems in current spatially smooth subspace face recognition methods. Using the penalized subspace learning framework, we then proposed two novel penalty functions, Laplacian of Gaussian (LOG) and Derivative of Gaussian (DOG), for subspace face recognition. LOG and DOG penalties introduce a scale parameter, and thus are more flexible in controlling the degree of smoothness. Experimental results indicate that the proposed methods are effective for face recognition, and achieve higher recognition accuracy than the original subspace methods.
Description: 6th International Symposium on Neural Networks, ISNN 2009, Wuhan, 26-29 May 2009
ISBN: 3642015123
ISSN: 0302-9743
EISSN: 1611-3349
DOI: 10.1007/978-3-642-01513-7_48
Appears in Collections:Conference Paper

View full-text via PolyU eLinks SFX Query
Show full item record


Last Week
Last month
Citations as of Apr 6, 2019

Page view(s)

Last Week
Last month
Citations as of Oct 22, 2019

Google ScholarTM



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.