Study on brain-computer interface based on mental tasks

In this paper, a novel method was proposed, which could realize brain-computer interface by means of distinguishing two different imaginary tasks of relaxation-meditation and tension-imagination based on Electroencephalogram (EEG) signal. When subjects performed the task of relaxation-meditation or tension-imagination, the output EEG signals of the subjects from the central parieto-occipital region of PZ electrode were recorded by the digital EEG device. By means of drawing Hilbert time-frequency amplitude spectrum and selecting the statistical properties of amplitude within different time-frequency bands as characteristic vector set, then carrying out feature selection based on Fisher distance criterion, choosing former several elements of larger Fisher index to be multidimensional feature vector and at last inputting the eigenvector to Fisher classifier, and so brain-computer interface was realized. The experiment results of 15 volunteers showed that the average of classification correct ratio was 90.3% and the highest was 95%. Due to only one electrode adopted, if some coding way was adopted, the brain-computer interface technology could be more easily used in robot control.