A Novel Discriminant Non-Negative Matrix Factorization Algorithm With Applications to Facial Image Characterization Problems

Author

Kotsia, Irene ; Zafeiriou, Stefanos ; Pitas, Ioannis

Author_Institution

Dept. of Inf., Aristotle Univ. of Thessaloniki, Thessaloniki

Volume

2

Issue

3

fYear

2007

Firstpage

588

Lastpage

595

Abstract

The methods introduced so far regarding discriminant non-negative matrix factorization (DNMF) do not guarantee convergence to a stationary limit point. In order to remedy this limitation, a novel DNMF method is presented that uses projected gradients. The proposed algorithm employs some extra modifications that make the method more suitable for classification tasks. The usefulness of the proposed technique to frontal face verification and facial expression recognition problems is demonstrated.

Keywords

face recognition; gradient methods; image classification; matrix decomposition; discriminant non negative matrix factorization; facial expression recognition; facial image characterization; frontal face verification; image classification; linear discriminant analysis; projected gradients; Convergence; Eyebrows; Eyes; Face recognition; Facial features; Image databases; Image recognition; Informatics; Linear discriminant analysis; Principal component analysis; Facial expression recognition; frontal face verification; linear discriminant analysis; non-negative matrix factorization (NMF); projected gradients;

fLanguage

English

Journal_Title

Information Forensics and Security, IEEE Transactions on

Publisher

ieee

ISSN

1556-6013

Type

jour

DOI

10.1109/TIFS.2007.902017

Filename

4291552