Motion Analysis of Visual Line in Eye-gaze-based Character Input Process Using an Onscreen Keyboard

Abstract

Assistive devices that use eye-gaze-based methods of data input are becoming increasingly popular, especially among individuals with severe physical disabilities. Current onscreen keyboards, however, are simple applications that struggle to accurately gather desired input-data or filter out information that is mere “noise” resulting from physiological eye movement. The present work describes our attempts to clarify the focal point while hovering over a key, the duration of the hover and the position of the key relative to the target key in case of using a popular gaze input, as a subject of basic research for interface design suited to eye-gaze input. We used two keys, a circle and a square, and categorized eye gaze into 3 different types ; short- and long-fixation, and character entry, according to its duration. Then the incidence of each was computed. Empirical results revealed that long fixation and character entry were observed on the target- and surrounding keys, while short fixation was observed on peripheral keys. As for the circular key, long fixation and character entry tended to occur in a relatively small region surrounding the target key.