A novel gaze tracking method based on the generation of virtual calibration points

Area of Science:

• Human-Computer Interaction
• Biomedical Engineering
• Computer Vision

Background:

• Conventional gaze-tracking systems demand extensive user calibration.
• This inconvenience limits the practicality and user-friendliness of existing eye-tracking technologies.

Purpose of the Study:

• To develop an efficient and accurate gaze-tracking system with minimal calibration requirements.
• To enhance user experience by reducing the number of calibration points needed.

Main Methods:

• A novel gaze-tracking system utilizing a large screen and a lightweight user device.
• Initial calibration involves noting four pupil centers, with five virtual pupil centers generated via a multilayer perceptron.
• Multi-geometric transforms are applied using actual and virtual pupil centers to calculate gaze position, accounting for nonlinear eye movements.

Main Results:

• The proposed method demonstrates superior accuracy compared to existing gaze-tracking techniques.
• The system effectively handles the distorted quadrangle formed by pupil centers due to nonlinear eyeball movement.
• Reduced gaze-detection error is achieved by employing multi-geometric transforms.

Conclusions:

• The new gaze-tracking method offers a more convenient and accurate alternative to conventional systems.
• This approach significantly improves the precision of eye-tracking by addressing nonlinearities in human eye movement.
• The findings suggest a promising advancement for applications requiring precise gaze detection with simplified calibration.