Activity related biometric authentication using Spherical Harmonics

Anastasios Drosou, Konstantinos Moustakas, Dimosthenis Ioannidis, Dimitrios Tzovaras

Abstract: This paper proposes a novel activity-related authentication method for ambient intelligence environments, based on Spherical Harmonics Analysis (SHA). The approach is based on the representation of the user's behavioural response with an Activity Surface (AS), while performing a regular office activity. Multi-referenced Spherical Harmonics Analysis has been utilized for extracting the corresponding Spherical Harmonics Coefficients (SHC)from each surface, which stand for the user's biometric signature. The experimental scenario has been carried out on a database that consists of 29 persons performing a phone-conversation activity. The authentication potential of the proposed biometric features has been seen to be significantly high in the performed experiments, while it is enhanced with recording view-angle invariance.