WCSE

Feature Extraction and Classification of Four-Class Motor Imagery Signals Based on LCD and CSP

Abstract— The common spatial pattern (CSP) can effectively extract the spatial information of motor imagery (MI) signals, but ignores time and frequency domain information of EEG signals. In order to overcome this problem, a new method is proposed in this paper, which combines the CSP method with the time-frequency analysis method Local Characteristic-scale Decomposition (LCD) to extract the timefrequency information and improve the classification accuracy. The effectiveness of the algorithm was verified by conducting experiments with the BCI competition dataset. The results show that the proposed method improves the recognition rate of MI signals, and has potential for the application of portable BCI systems in rehabilitation field.

Index Terms— brain-computer interface, motor imagery, common spatial pattern, local characteristic-scale decomposition, time-frequency analysis

Qingsong Ai, Yi Xie, Kun Chen
School of Information Engineering, Wuhan University of Technology, CHINA
Qingsong Ai, Kun Chen
Key Laboratory of Fiber Optic Sensing Technology and Information Processing (Wuhan University of Technology), Ministry of Education, CHINA

[Download]

Cite: Qingsong Ai, Yi Xie, Kun Chen, "Feature Extraction and Classification of Four-Class Motor Imagery Signals Based on LCD and CSP," Proceedings of 2017 the 7th International Workshop on Computer Science and Engineering, pp. 711-716, Beijing, 25-27 June, 2017.