APPARATUS AND METHOD FOR MOTOR IMAGERY CLASSIFICATION USING ADAPTIVE NOISE CANCELLATION AND CORRELATION OPTIMIZATION

摘要

Disclosed is a method for motor imagery classification which improves classification accuracy. The method for motor imagery classification is performed by an apparatus for motor imagery classification and comprises: (a) a step of applying an adaptive noise cancellation algorithm to remove noise of electroencephalogram signals measured from a plurality of electrodes for each class to be classified and obtain preprocessed electroencephalogram signals; (b) a step of using a rotation matrix having a rotation angle () to rotate the preprocessed electroencephalogram signals to obtain rotation electroencephalogram signals; (c) a step of calculating correlation coefficients between the rotation electroencephalogram signals by the rotation angle (); (d) a step of calculating Fishers ratio for the correlation coefficients between the rotation electroencephalogram signals; (e) a step of optimizing the calculated correlation coefficients between the rotation electroencephalogram signals to obtain optimum correlation coefficients; (f) a step of extracting n optimum correlation coefficients in descending order by Fishers ratio among the optimum correlation coefficients as a feature vector; and (g) a step of using the feature vector to learn a classifier, wherein n is a natural number.;COPYRIGHT KIPO 2020