Exploring the Potential of Event Camera Imaging for Advancing Remote Pupil-Tracking Techniques

Abstract

Pupil tracking plays a crucial role in various applications, including human–computer interactions, biometric identification, and Autostereoscopic three-dimensional (3D) displays, such as augmented reality (AR) 3D head-up displays (HUDs). This study aims to explore and compare advancements in pupil-tracking techniques using event camera imaging. Event cameras, also known as neuromorphic cameras, offer unique benefits, such as high temporal resolution and low latency, making them well-suited for capturing fast eye movements. For our research, we selected fast classical machine-learning-based computer vision techniques to develop our remote pupil tracking using event camera images. Our proposed pupil tracker combines local binary-pattern-features-based eye–nose detection with the supervised-descent-method-based eye-nose alignment. We evaluate the performance of event-camera-based techniques in comparison to traditional frame-based approaches to assess their accuracy, robustness, and potential for real-time applications. Consequently, our event-camera-based pupil-tracking method achieved a detection accuracy of 98.1% and a tracking accuracy (pupil precision < 10 mm) of 80.9%. The findings of this study contribute to the field of pupil tracking by providing insights into the strengths and limitations of event camera imaging for accurate and efficient eye tracking.

Keywords

• eye tracking
• eye detection
• event camera
• dynamic vision sensor
• system latency
• Autostereoscopic 3D display