Publication details

Home Publications Publication details

Transparent Facial Recognition for Mobile Devices
Clarke NL, Karatzouni S, Furnell SM
Proceedings of the 7th Security Conference, Las Vegas, USA, 2nd-3rd June, 2008
Download links:  Download PDF

The growing popularity and increasing functionality of mobile devices has resulted in them becoming increasingly desirable targets for theft and unauthorised use. Key to ensuring effective protection is providing an effective and usable form of user authentication. Traditional approaches, such as the PIN, have been shown to be ineffective and inconvenient. Indeed, the intrusive nature of this form of authentication is itself a barrier to acceptance by users. This paper discusses the application of alternative authentication technologies, such as biometrics in a transparent and continuous fashion – removing the inconvenience of intrusive authentication and extending the identity verification beyond point-of-entry. Unfortunately, for this type of authentication to operate, the individual biometric techniques need to be able to operate in a transparent fashion, where the system is unable to control key factors that are required for successful operation. For instance, facial recognition performs very well when facial orientation, illumination, distance and camera quality can all be controlled. Under a transparent mode of operation, these factors cannot be so closely controlled. The paper presents a study aimed at validating the feasibility of improving facial recognition algorithms for use transparently by the means of a composite facial template. The study specifically addresses the performance issues of utilising facial images with varying facial orientations, and has shown that the usability of facial recognition algorithms can be significantly improved through the use of this composite template at a minimal expense to the level of security being provided.

Clarke NL, Karatzouni S, Furnell SM