Paper by Alexandru Popovici, Theerawit Wilaiprasitporn, Tohru Yagi.
Abstract—Electrooculography (EOG) enables users to use specific eye movements as inputs for various applications, without using their fingers. However, online classification of such signals often requires long or sophisticated calibration procedures and multiple electrodes, which makes the resulting systems not practical for everyday use. To this end, a single-channel bio-potential recording system was used to develop a Personal Identification Number (PIN) input application, based on a simple and short unsupervised calibration method. The average accuracy achieved from five healthy subjects was 98.24%. The high accuracy, combined with the use of single-channel recording with convenient electrode placement resulted in a system that could be embedded in a reliable wearable human-computer interface (HCI) device.
Link to paper: