Compressed phase component for low resolution face recognition
by Naser Zaeri
International Journal of Biometrics (IJBM), Vol. 6, No. 4, 2014

Abstract: Most of the current algorithms for face recognition do not consider images of low resolution that commonly exist in real life applications. In this paper, we address this issue and propose an efficient solution for face recognition for systems that deal with such images and utilise limited storage. The new method transforms the input data into a different form that identifies the face image database structure, for which certain data can be dropped (or compressed) without the fear of performance deterioration. It implements a lossy compression scheme on the discriminant Fourier phases of the face image components. A thorough study and comprehensive experiments relating to time consumption and computational complexity versus system performance are applied to benchmark face image databases. It will be shown that the proposed method offers exceptional performance and, at the same time, achieves substantial savings in computational time when compared to other known methods. The experimental results reveal that a recognition rate higher than 98% is achieved at a compression ratio of 1.78, with training time less than four minutes for a database consisting of 2,360 images.

Online publication date: Sat, 07-Feb-2015

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Biometrics (IJBM):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com