Text-entry interface using continuous eye gestures from EOG

Type: Semester thesis
Students: Stefan Dahinden, Stephan Lütolf
Advisor: Andreas Bulling, Clemens Lombriser
Project: Eye Movements and Visual Perception

This thesis investigated the design and implementation of a continuous text-entry interface using eye movements. In contrast to common systems using video cameras, the interface was supposed to be controlled using electrooculography (EOG), a measurement technique that uses body-worn sensors to track eye movements and allows to detect eye gestures.

In the first step, an existing fast text-entry interface, mainly used for mouse input called "Dasher" was analyzed to identify the main causes making it fast and comfortable to use. Additionally, different eye based text-entry interfaces were analyzed to investigate the requirements for designing a system for eye based human-computer interaction. Finally, an interface was developed and implemented that combines the findings gained from both evaluations.

In a user study with eight subjects the usability of the developed interface was evaluated and the performance was measured. The results were compared to other interfaces using different technologies and were discussed. The work showed that it is possible to adopt the "flow" experienced in Dasher while still using discrete inputs detected from EOG signals.