METHOD FOR ANALYZING CORRELATION CHARACTERISTICS OF ELECTROENCEPHALOGRAM BETWEEN EEG CHANNELS AND METHOD FOR RECOGNIZING HUMAN INTENTIONS USING CORRELATION CHARACTERISTICS OF ELECTROENCEPHALOGRAM BETWEEN EEG CHANNELS

摘要

A method for recognizing a human intention using inter-channel correlation for electroencephalograms measured from multiple channels according to an embodiment of the present invention comprises the steps of: (A1) measuring the electroencephalograms of a subject from M channels and computing inter-channel correlation coefficients for the electroencephalograms; (A2) computing p statistical feature values for the computed inter-channel correlation coefficients; (A3) selecting w inter-channel correlation coefficients among the computed inter-channel correlation coefficients in the order that a positive value is larger and selecting w inter-channel correlation coefficients among the computed inter-channel correlation coefficients in the order that a negative value is larger to extract reference channels; and (A4) selecting, as the intention of the subject, a candidate model, in which statistical feature values for inter-channel correlation coefficients at the same locations as the reference channels are closest to statistical feature values for inter-reference channel correlation coefficients, among candidate models. By providing inter-channel correlation characteristic information for electroencephalograms, electroencephalograms according to various brain activities can be analyzed and recognized. [Reference numerals] (A1) Measure the electroencephalograms of a subject from M channels and compute inter-channel correlation coefficients for the electroencephalograms; (A2) Compute p statistical feature values for the computed inter-channel correlation coefficients; (A3) Select w inter-channel correlation coefficients among the computed inter-channel correlation coefficients in the order that a positive value is larger and select w inter-channel correlation coefficients among the computed inter-channel correlation coefficients in the order that a negative value is larger to extract reference channels; (A4) Select, as the intention of the subject, a candidate model, in which statistical feature values for inter-channel correlation coefficients at the same locations as the reference channels are closest to statistical feature values for inter-reference channel correlation coefficients, among candidate models; (AA) Start; (BB) End