A Novel Remote Eye Gaze Tracking System Using Line Illumination Sources

This paper proposes a novel system to estimate the 3D point of gaze using observations of the pupil center and corneal reflections (glints). A mathematical model is developed with two solutions to estimate the corneal center efficiently using lines of LED lights. Differing from existing 3D approaches requiring associating light sources with glints, the model automatically associates glints and LED lines and can handle missing glints well. The new model also enables a user-friendly calibration process, allowing natural head movement. Experiments demonstrate that the proposed system can achieve accurate gaze estimation with natural head movement. The performance is impressive when using the natural calibration, requiring less user cooperation.