earticle

영어

Conventional authentication systems use secret knowledge like password either from alphanumeric PIN to graphical click-based or pattern password that impose memory burden to users. Biometrics appears to answer the problem related to conventional system. We propose an ear biometrics as an alternative to popular facial biometrics. One of the ways to implement biometrics authentication system is by authenticating them via image or video captured using a dedicated terminal as biometrics enrolment module. This biometrics module is pricey thus adding cost to overall cost of having smart properties for people. In addition, it can be destroyed by thieves to bypass biometrics authentication after alarm system being turned off. We perceive that smartphone camera can be used as replacement of dedicated enrolment module instead. The replacement will result a biometrics enrolment terminal that is firstly mobile and automatically practical unlocking home while user is within home proximity. With location-based service (LBS) support, it can enable convenient and implicit owner authentication system when accessing the protected smart premise and property (like smart home). In a situation that user calls to home then LBS could detect user's location within close range of proximity thus authenticating a user via its front camera that faces to ear while being used on calling home. Furthermore, it is cost-efficient because eliminating the necessity to install dedicated enrolment terminal in a property like smart home. In this paper, we present a novel approach to ear biometrics which considers both shape and texture information to represent ear image during ear recognition computation for authentication. We use local invariant patterns as ear image descriptor during recognition to have lightweight but accurate ear biometrics system on securing smart homes. We improve the original local invariant against noise that may reduce accuracy in potential deterrent conditions like over noise and illumination changes.