A Novel Cross-Ratio Based Gaze Estimation Method

Eye movement is the most natural expression of human thoughts, therefore it plays an important role in human-computer interaction. Over the past few decades, researchers have been pursuing for high-accuracy, easy-to-use eye-tracking techniques. The gaze tracking technology based on cross-ratio is a hot topic in the field due to its advantages of simple hardware device and allowing free head movement. Given the complexity of the user calibration procedure and the error caused by the coplanar assumption in the cross-ratio based method, we propose a new method to approximate the pupil plane step by step through constructing the front and rear planes, and then estimate the point-of-regard based on cross-ratio invariance. Simulation results show that the proposed method not only simplifies the system configuration and user calibration procedure, but also avoids the error caused by the non-coplanarity between the corneal surface reflection points and the pupil center.