Research on Multi-feature Extraction Method in EEG Signal of Motor Imagination

Abstract

The feature extraction of EEG signals of motor imaging is a very important step in the brain-computer interface system. We usually use the event-related synchronization/desynchronization (ERS/ERD) phenomenon of EEG signals, and use common motor imagination EEG signal’s feature extraction processing algorithms, including Common Spatial Pattern (CSP), Power Spectral Density (PSD), Autoregressive model (AR), Discrete Wavelet Transform DWT) method and other methods to extract features of EEG signals. This paper expounds the basic principle of the algorithm, and finally uses the support vector machine (SVM) to classify the left-hand and right-hand motor imagery patterns after processing the four methods. Comparative experiments show that the feature extraction method of multi-feature fusion better represents EEG features, and a better classification effect can be achieved by using SUM classification. This study can meet the requirements of BCI for high recognition rate.