A Repeated Bisection CSP Feature Extraction Algorithm of Four-Class Motor Imagery EEG

In this paper, based on analysis of the phenomenon of ERD/ERS in brain-computer interface (BCI), an improved repeated bisection common spatial pattern (RB-CSP) algorithm was presented to extract the features of four-class motor imagery EEG and the support vector machine (SVM) was used to classify. The experimental results show that, the proposed algorithm can reduce time consumption and complexity, can produce high classification accuracy, compared with OVR-CSP of the CSP traditional extensions. The proposed algorithm provides a new solution to real-time BCI systems.