EEG ANALYSIS FOR DIGIT RECOGNITION BY TACTILE AND VTBROTACTILE STIMULATIONS

摘要

Artificial rehabilitative aids to enable object recognition to the disabled as well as robot aided and telenavigating systems require sending feedback signals to the human operator to enable accurate control. This work is a preliminary step towards the development of such systems using a Brain Computer Interface. In this work Electroencephalography (EEG) responses to tactile and vibrotactile stimulations, as alternate sensory means than vision, for recognizing ten digits, 0 to 9, has been studied. During tactile stimulation subjects are instructed to palpate digits embossed on plain surfaces, while vibrotactile stimulus is provided by vibrating motors attached to the subjects' skin over their clothing in specific patterns resembling the seven segment display of digits. EEG analysis involves feature extraction and classification into the respective digit classes. Correlation between the EEG features from the two stimulations is investigated and a nonlinear correlation is found to exist between them. A maximum classification accuracy of 73.17%, average over ten digit classes and all subjects under experimentation is observed for vibrotactually stimulated EEG analysis.