Improvement of Eye Gesture Interface

Human-computer interaction is an important field in computer science. One means of such interaction is an eye gaze interface using both optical and electrical methods. Conventional optical devices for measuring eye movement are generally expensive and restrict user head movement. In this paper, we propose a novel eye gesture input interface (EGII) that does not require a head-mounted display. Precise eyemovement in combinations of direction is determined by electrooculographic signals amplified via AC coupling. We also developed and evaluated direction algorithms for determining features of the novel EGII. The best algorithm showed a success rate of 97% in recognizing eye movement in different direction combinations. We also created a small portable measurement device containing a biological amplifier to amplify signals and an A/D board for importing signals. Mean accuracy for the EGII was 95.6% out of 12 possible choices using the selected algorithm and our device. This device can thus be used anywhere and anyone.