An analysis of the electrooculography slow-oscillation during light adaptation under varying lighting intensities

Abstract

Electrooculography (EOG) is a cost-effective and non-invasive technique for measuring eye movements, widely used in clinical diagnostics and for eye-gaze estimation used to develop eye movement-based human-computer interface (HCI) systems. While EOG signals are often considered insensitive to illumination changes in the EOG-based HCI system domain, a standard clinical ophthalmic test indicates that the EOG signal amplitude for the same ocular displacement oscillates upon both light and dark onset. Hence, this study investigates the EOG sensitivity with the aim of clarifying any inconsistencies in the literature specifically for HCI applications. A recording protocol was designed to capture the behaviour of EOG signals under two typical indoor illumination levels (150 lx and 500 lx) for comparative analysis. Results obtained from four participants confirm that while steady-state sensitivity values are comparable between illumination levels, significant fluctuations after light onset may lead to over- or under-estimation of the user’s gaze, adversely impacting the gaze estimation accuracy. These findings suggest that while EOG-based HCI systems are largely insensitive to illumination levels under steady-state conditions, lighting transitions necessitate frequent recalibration to maintain accuracy. This work thus highlights the need to model this oscillatory process, enabling real-time compensation and minimising the need for recalibration in real-world EOG applications, and lays the groundwork for future studies.