A Stochastic Nonlinear Dynamical System for Smoothing Noisy Eye Gaze Data

2025-04-17

AI Summary

Eye movement tracking suffers from gaze point noise caused by hardware limitations, calibration drift, illumination variations, and blinks. To address this, we propose a stochastic nonlinear dynamical modeling framework tailored to real-world eye-tracking data. Our approach formulates a state-space model that explicitly captures the nonlinear evolution of gaze trajectories and inherent stochastic disturbances. We systematically analyze the coupling between process noise covariance and dynamic parameters in the Extended Kalman Filter (EKF), enabling joint optimization of robustness and responsiveness. Experiments demonstrate that our method significantly reduces gaze noise (average RMSE reduction of 38.2%), improves trajectory fitting accuracy, and yields outputs highly consistent with ground-truth eye movement recordings. This work establishes an interpretable, dynamics-based paradigm for high-precision, low-latency gaze estimation, with direct applicability in human-computer interaction and cognitive neuroscience experiments.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Motion & Tracking
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements

Abstract

In this study, we address the challenges associated with accurately determining gaze location on a screen, which is often compromised by noise from factors such as eye tracker limitations, calibration drift, ambient lighting changes, and eye blinks. We propose the use of an extended Kalman filter (EKF) to smooth the gaze data collected during eye-tracking experiments, and systematically explore the interaction of different system parameters. Our results demonstrate that the EKF significantly reduces noise, leading to a marked improvement in tracking accuracy. Furthermore, we show that our proposed stochastic nonlinear dynamical model aligns well with real experimental data and holds promise for applications in related fields.

Problem

• Smoothing noisy eye gaze data from trackers
• Reducing noise using extended Kalman filter
• Improving gaze tracking accuracy dynamically

Innovation

• Extended Kalman filter for gaze smoothing
• Stochastic nonlinear dynamical model
• Systematic parameter interaction exploration