A drag-and-drop type human computer interaction technique based on electrooculogram

Abstract

A fundamental limitation of human-computer interaction using electrooculogram (EOG) is a low accuracy of eye tracking performance and the head movement that violates the calibration of the on-monitor gaze coordinates. In this paper, we develop a drag-and-drop type interface with the EOG that can avoid a direct estimation of gaze location and can make users free from the restriction of head movement. To drag a cursor on the screen, the proposed system models the relationship between the amount of eye movement and the EOG amplitude with linear regression. Five subjects participated in the experiment to compare the proposed drag-and-drop type and the conventional direct gaze type interfaces. Performance measures such as efficiency and satisfaction showed the advantage of the proposed method with significant differences (p < 0.05).