Novel eye gaze tracking techniques under natural head movement.

Area of Science:

• Human-Computer Interaction
• Biomedical Engineering
• Computer Vision

Background:

• Existing remote eye gaze trackers require individual calibration and restrict head movement, limiting their use as computer input devices.
• Current systems necessitate frequent recalibration for each user, hindering seamless interaction.
• Unnatural head stabilization is often required, compromising user comfort and natural interaction.

Purpose of the Study:

• To develop novel eye gaze tracking solutions that enable natural head movement.
• To minimize the calibration procedure to a single session per user.
• To enhance the usability of eye gaze tracking technology for human-computer interaction.

Main Methods:

• A convex mirror model of the cornea is used to directly estimate 3-D eye gaze, determining the visual axis from the optic axis.
• An implicit gaze mapping function estimates gaze points without direct 3-D gaze calculation.
• A dynamic computational head compensation model continuously updates the gaze mapping function to accommodate head movements.

Main Results:

• The proposed techniques accurately estimate eye gaze under natural head movement.
• Calibration is reduced to a single, one-time procedure for each individual.
• The methods overcome the limitations of traditional eye trackers regarding head motion tolerance and recalibration frequency.

Conclusions:

• The developed eye gaze tracking methods significantly improve usability for natural computer input.
• These advancements represent a crucial step towards the acceptance of eye trackers as intuitive input devices.
• The research addresses key challenges in remote eye gaze tracking, paving the way for broader adoption.