Gaze Tracking with Head Pose Estimation and Compensation

The visible-spectrum gaze tracking system uses a single high-speed color camera, eliminating the need for an infrared light source or sensor. While this setup significantly enhances the user experience, compensating for head movement becomes more challenging. This paper proposes a new gaze tracking system incorporating a head pose estimation tool to identify facial feature points and solve the 2D-to-3D correspondence problem. The 3D coordinates of these feature points are then used to estimate head movement and calculate the initial location of the iris regions. Experimental results show that the accuracy and precision of the gaze tracking system are improved when users move and rotate their heads.