Multispectral face liveness detection method based on gradient features

Ya-Li Hou; Xiaoli Hao; Yueyang Wang; Changqing Guo

doi:10.1117/1.OE.52.11.113102

7 November 2013 Multispectral face liveness detection method based on gradient features

Ya-Li Hou, Xiaoli Hao, Yueyang Wang, Changqing Guo

Author Affiliations +

Optical Engineering, Vol. 52, Issue 11, 113102 (November 2013). https://doi.org/10.1117/1.OE.52.11.113102

Abstract

Face liveness detection aims to distinguish genuine faces from disguised faces. Most previous works under visible light focus on classification of genuine faces and planar photos or videos. To handle the three-dimensional (3-D) disguised faces, liveness detection based on multispectral images has been shown to be an effective choice. In this paper, a gradient-based multispectral method has been proposed for face liveness detection. Three feature vectors are developed to reduce the influence of varying illuminations. The reflectance-based feature achieves the best performance, which has a true positive rate of 98.3% and a true negative rate of 98.7%. The developed methods are also tested on individual bands to provide a clue for band selection in the imaging system. Preliminary results on different face orientations are also shown. The contributions of this paper are threefold. First, a gradient-based multispectral method has been proposed for liveness detection, which considers the reflectance properties of all the distinctive regions in a face. Second, three illumination-robust features are studied based on a dataset with two-dimensional planar photos, 3-D mannequins, and masks. Finally, the performance of the method on different spectral bands and face orientations is also shown in the evaluations.

Citation Download Citation

Ya-Li Hou, Xiaoli Hao, Yueyang Wang, and Changqing Guo "Multispectral face liveness detection method based on gradient features," Optical Engineering 52(11), 113102 (7 November 2013). https://doi.org/10.1117/1.OE.52.11.113102

Published: 7 November 2013

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available