Polarization-resolved imaging improves eye tracking

2025-11-06

AI Summary

To address degraded gaze estimation accuracy caused by eyelid occlusion, inter-pupillary distance variation, and pupil size changes, this paper proposes Polarization-Enhanced Tracking (PET), a novel eye-tracking system. PET uniquely integrates linearly polarized near-infrared illumination, a polarization-filter array camera, and a convolutional neural network. Leveraging the distinct polarization-dependent reflectance properties of scleral and corneal tissues, PET extracts structured features invisible in conventional intensity images, enabling robust gaze estimation. Evaluated on a large-scale dataset comprising 346 participants, PET reduces the median 95%-ile absolute gaze estimation error by 10–16% over intensity-based baseline models across diverse challenging conditions—including partial occlusion and anatomical variability. This work constitutes the first systematic demonstration of the modeling value and practical viability of tissue-level polarization effects for gaze sensing.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Large-scale security measurements
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Polarization-resolved near-infrared imaging adds a useful optical contrast mechanism to eye tracking by measuring the polarization state of light reflected by ocular tissues in addition to its intensity. In this paper we demonstrate how this contrast can be used to enable eye tracking. Specifically, we demonstrate that a polarization-enabled eye tracking (PET) system composed of a polarization--filter--array camera paired with a linearly polarized near-infrared illuminator can reveal trackable features across the sclera and gaze-informative patterns on the cornea, largely absent in intensity-only images. Across a cohort of 346 participants, convolutional neural network based machine learning models trained on data from PET reduced the median 95th-percentile absolute gaze error by 10--16% relative to capacity-matched intensity baselines under nominal conditions and in the presence of eyelid occlusions, eye-relief changes, and pupil-size variation. These results link light--tissue polarization effects to practical gains in human--computer interaction and position PET as a simple, robust sensing modality for future wearable devices.

Problem

• Improving eye tracking accuracy using polarization-resolved near-infrared imaging
• Revealing trackable ocular features invisible in intensity-only images
• Reducing gaze estimation errors under various challenging eye conditions

Innovation

• Polarization-filter-array camera captures eye features
• Linearly polarized NIR illuminator enhances contrast
• Convolutional neural network reduces gaze tracking error