基于CSP与SVM算法运动想象脑电信号分类.pdf

第31 卷第8期 东 北 大 学 学 报 ( 自 然 科 学 版 ) Vol31, No. 8 2 0 10 年 8 月 Journal of Northeastern U niversit ( Natural Science) Aug. 2 0 1 0 CSPSVM 刘 冲, 赵海滨, 李春胜, 王 宏 ( , 110004) : ( ) , ( common spatial pattern, CSP) BCI 2003 ; , CSP C3, Cz C4 , 8286% , 409 s, 047 bit, 0431bit/ s BCI 2003 , , BCI : ; ; ; ; ; : R 318 : A : 20 10) CSP/SVMBased EEG Classification of Imagined Hand Movements LI U Chong , ZHA O H aibin, LI Chunsheng, WANG H ong ( School of Mechanical Engineering Automation, Northeastern Universit , Shen ang 110004, China. Correspondent: LIU Chong, Email: congliu @ me. neu. edu. cn) Abstract: For the BCI ( braincomputer interface) to classif the different imagined movements of both left and right hands, the method of CSP ( common spatial pattern) w as used to extract the features of BCI 2003 competitive dataset . T hen, the CSP based on a sliding time w indow was used to filter the EEG( electroencephalogram) data from the electrodes C3, Cz and C4, with the SVM ( support vector machine) used as a classifier of the features. As a result, the highest accurac of classification is 82. 86% w ith the best classification time point 4. 09 sec, max imum mutual information ( M I) 0. 47 bit and maximum M I steepness 0. 431 bit/ s. Compared to the results of BCI 2003 competition, the method as above can provide greatl improved max imum M I steepness, thus verif ing that the method is more adaptable to w hat are required b the online BCI s stem. Key ords: EEG; BCI ( braincomputer interface) ; CSP ( common spatial pattern ) ; SVM ( support vector machine) ; mutual information; time for classification ( braincomputer inter