Exploiting symmetry in two-dimensional clustering-based discriminant analysis for face recognition

Ioannis Pitas, Konstantinos Papachristou, Anastasios Tefas

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

210 Downloads (Pure)

Abstract

Subspace learning techniques are among the most popular methods for face recognition. In this paper, we propose a novel face recognition technique for two dimensional subspace learning which is able to exploit the symmetry nature
of human faces. We extent the Two Dimensional Clustering based Discriminant Analysis (2DCDA) by incorporating an appropriate symmetry regularizer into its objective function in order to determine symmetric projection vectors. The
proposed Symmetric Two Dimensional Clustering based Discriminant Analysis technique has been applied to the face recognition problem. Experimental results showed that the proposed technique achieves better classification performance
in comparison to the standard one.
Original languageEnglish
Title of host publication2015 23rd European Signal Processing Conference (EUSIPCO)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Pages155-159
Number of pages5
ISBN (Electronic)9780992862633
ISBN (Print)9781479988518
DOIs
Publication statusPublished - 28 Dec 2015
Event23rd European Signal Processing Conference, EUSIPCO 2015 - Nice, France
Duration: 31 Aug 20154 Sep 2015

Publication series

NameProceedings of the European Signal Processing Conference (EUSIPCO)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
ISSN (Print)2219-5491

Conference

Conference23rd European Signal Processing Conference, EUSIPCO 2015
CountryFrance
CityNice
Period31/08/154/09/15

Keywords

  • face recognition
  • subspace learning
  • symmetry regularizer
  • two-dimentional clustering-based discriminant analysis

Fingerprint

Dive into the research topics of 'Exploiting symmetry in two-dimensional clustering-based discriminant analysis for face recognition'. Together they form a unique fingerprint.

Cite this