Biometric Identification via Oculomotor System Based on the Volterra Model

Abstract

In recent years, there has been an increase in interest in biometrics research involving the use of brain characteristics commonly known as behavioral traits. Human eyes contain a rich source of idiosyncratic information which may be used for the recognition of an individual's identity. This article implements an innovative experiment and a new approach to processing human eye movements, ultimately aimed at biometric identification of individuals. In our experiment, the subjects observe special test visual stimuli, which are generated on the computer monitor screen. The eye movements are tracked in dynamics providing information for constructing a nonparametric nonlinear dynamic model (Volterra model) of a human's oculomotor system (OMS) in the form of multivariate transient functions. The implemented method treats eye trajectories as 2-D distributions of points on the “Coordinate-Time” plane. The efficiency of dynamic characteristics for personality identification is confirmed by examples of models built on the basis of data from real experiments. The resulting OMS models are a source of information for the selection of informative features, in the space of which the decisive rule of optimal identification of individuals is determined using machine learning methods. Promising results at the task of identification according to behavioral characteristics of an individual have been obtained - recognition accuracy is higher than 97%.