A New Eye-tracking Method with Image Feature Based Model for Mobile Devices.

In order to improve the computational efficiency and accuracy of eye-tracking methods for mobile devices, this paper proposes an eye-tracking method based on face features and head pose estimation. The method first uses image processing to detect human eye feature points, and then establishes an eyepoint mapping model based on Gaussian process regression; on this basis, the head pose estimation is combined to compensate for the eye-point coordinates; finally, a prototype eye-tracking system based on mobile devices is designed and developed. The experimental results show that the system has high accuracy and robustness, thus verifying the feasibility and effectiveness of the method in this paper.