A novel approach to 3-D gaze tracking using stereo cameras

Area of Science:

• Computer Vision
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Traditional 3-D gaze tracking often requires complex and error-prone user-dependent calibration procedures.
• The need for efficient and accurate eye-tracking systems is growing in various fields, including virtual reality and assistive technologies.

Purpose of the Study:

• To develop a novel, user-independent 3-D gaze tracking system.
• To eliminate the need for inconvenient system calibration processes.
• To achieve real-time gaze tracking with high accuracy.

Main Methods:

• Utilized multiple cameras and multiple point light sources for 3-D computer vision-based eye-axis estimation.
• Developed a real-time system capable of 30 gaze measurements per second.
• Implemented a simple, rapid calibration method requiring users to fixate on a target for 2-3 seconds.

Main Results:

• The developed 3-D gaze tracking system achieved an average estimation error of less than 1 degree across six subjects.
• The system successfully eliminated the need for user-dependent calibration, simplifying setup and reducing potential errors.
• Real-time performance was demonstrated with 30 gaze measurements per second.

Conclusions:

• The proposed novel approach offers a highly accurate and user-friendly 3-D gaze tracking solution.
• The elimination of user-dependent calibration significantly enhances the practicality and efficiency of the system.
• This technology shows significant promise for real-time applications requiring precise eye-tracking capabilities.