Online Offset Correction of Remote Eye Tracking Data: A Novel Approach for Accurate Gaze-Based Mouse Cursor Control

Abstract

Abstract. Camera-based eye-and gaze tracking systems have a wide range of application areas including cognitive science, biometrics, usability studies and Assistive Technology. Although the accuracy of remote eye trackers improved considerably in recent years, it is still impractical to use eye tracking systems for cursor control because of jitter and inaccurate positioning. This work presents a novel approach for online offset correction of the estimated gaze point, so that state-of-the art, affordable remote eye tracking devices can be used for cursor control in desktop applications without special GUI adaption. For gaze point correction, additional sensor systems or input devices can be combined with the eye tracking device. In evaluations with test users it could be shown that the hit rate of small targets presented at random screen locations could be increased from 54% to 89% which significantly improved the usability of gaze tracking based mouse cursor control.