左右手运动想象脑电模式识别研究

如何提高左右手运动想象脑电信号的分类率是脑机接口研究领域的一个热点话题。基于美国EGI64导脑电采集系统得到3名健康被试的脑电数据,首先,采用独立成分分析（Independent Component Analysis,ICA）对采集的数据进行去噪处理;然后,利用离散小波变换方法对分解C3/C4处的EEG平均功率信号,选用尺度6上逼近系数A6的重构信号作为脑电特征信号;最后,用Fisher线性判别分析法（Fisher Linear Discriminant Analysis,FLDA）、支持向量机方法 （Support Vector Machines,SVM）和极限学习机分类方法 （Extreme Learning Machine,ELM）分别对特征信号进行分类。分类结果表明：极限学习机分类方法得出的平均分类率要高于Fisher方法与SVM方法的平均分类率,可以达到92%,而且运行速度也高于另两种分类算法。

Study of Discrimination between Left and Right Hand Movements Imagery Event--Related EEG Pattern

Recently, accurate classification of imaginary left and right hand movemetns of EEG is an important issue in brain0computer interface （BCI）. Based on EEG data of 3 subjects which collected by --American EGI 640--channel EEG colletion system, firstly, the effective de--noising processing to collected data by the independent component analysis method is carried out. Secondly, discrete wavelet transform meth od is used to decompose the average power of the channel C3/C4 in left and right hand movements image-ry. The reconstructed signal of approximation coefficient A6 on the sixth level is selected to build up feature signal. Finally, to classify the feature signal respectively by Fisher Linear Discriminant Analysis （FL- DA）, Support Vector Machines （SVM） and Extreme Learning Machine （ELM） methods. The classification results show that the average classification rate of --ELM is higher than that of FLDA and SVM, which can achieve 92%. The running speed of ELM is also faster than the other two methods.