According to the present invention, a motion imagination brain wave signal classification apparatus and a method thereof comprises: a step of receiving a brain wave signal obtained during preset time through a motion imagination experiment; a step of using a filter bank configured by uniformly dividing a frequency range into preset number to divide a frequency of the brain wave signal to correspond to the frequency range; a step of extracting and normalizing respective characteristics information by the frequency range of the brain wave signal through a normalization common section pattern (R-CSP) filter; a step of selecting a pair of the preset number of characteristics information among the extracted characteristics information using a Mutual Information Based Individual Feature (MIBIF) algorithm to configure a parameter set for an ensemble; and a step of classifying imagination motions of the brain wave signal by using an ensemble technique based on the selected pair of the characteristics information. According to the present invention, it is possible to improve performance by applying and ensembling the normalization common section pattern (R-CSP) filter to a brain wave passing through the filter bank to accurately classify characteristics of imagined motions.
展开▼
