Recognition of reading activity from the saccadic samples of electrooculography data

Abstract

In this paper, recognition of human activity, involving patterned eye movements, such as reading is introduced. Recognition of human activities is an important part in implementing ubiquitous, context-aware computer applications where computers can communicate with humans in a more interactive manner. Eye movement is a potential instrument in activity recognition as most of human activities involve movement of eyes. This paper describes a method to recognize reading activity from the eye movement patterns. Electrooculography (EOG) signal is used to quantify the eye movements in terms of electrostatic potential. The EOG signal is recorded using electrodes, placed at appropriate positions around the eyes. The extracted EOG signal is then analysed to detect eye movement patterns in connection to reading activity.