Head-Free, Remote Gaze Detection System Based on Pupil-Corneal Reflection Method with Using Two Video Cameras - One-Point and Nonlinear Calibrations

摘要

We developed a pupil-corneal reflection method-based gaze detection system, which allows head movements and achieves easy gaze calibration. The proposed gaze detection theory determines gaze points on a PC screen from the vector from the corneal reflection to pupil center, 3D pupil position, two cameras position, etc. In a gaze calibration procedure, after a user is asked to gaze one specific calibration target at a center of a PC screen, the nonlinear characteristic of the eyes has been automatically corrected while the user is using this gaze system. The experimental results show that the proposed calibration method improved the precision of gaze detection during browsing web pages. In addition, the average gaze error in the visual angle is less than 0.6 degree for the nine head positions.