Algorithm to detect six basic commands by the analysis of electroencephalographic and electrooculographic signals

Abstract

The Electroencephalographic signals are commonly used for developing brain-machine interfaces (BMI), in fact is the most used biological signal to translate brain's commands to the computer. Some additional physiological measures have been used along with EEG in order to obtain more robust and more accurate BMI systems. However, since very sophisticated recording devices are more available, signal processing is getting complicated, mainly due to the invested computational time in signal extraction and pattern recognition. Therefore, processing time in BMI could be too long, which is useless for some applications, for instance, devices used in rehabilitation engineering, or some robotic systems. In this paper, we propose a six commands recognition algorithm using only one EEG bipolar connection (O1-P3) in combination with bilateral electrooculographic signals. Our algorithm could identify these six commands based on simple temporal analysis with an average recognition accuracy of 97.1% for the selected sample of subjects. The average recognition time do not last more than 0.5 seconds after one of the events occurred.