摘要
通过对运动想象脑电信号的分类,对受试者进行身份识别。采用一种盲源分离算法——二阶盲辨识对运动想象脑电信号进行处理,提高运动想象脑电信号的信噪比,进而采用Fisher距离对处理后的信号进行特征提取,最后采用BP神经网络对特征集进行分类,从而实现对受试者的身份识别。对3位受试者的4类运动想象脑电信号分别进行了分类识别,结果显示,4类运动想象脑电信号的识别率均达到80%左右,其中最高的是想象舌动脑电信号,其识别率达到88.1%,这在类似研究中属于较高的水平。
Subjects are identified by classifying motor imagery EEG signal.Second-Order Blind Identification(SOBI),a Blind Source Separation(BSS) algorithm is applied to preprocess EEG data for higher signal-to-noise ratio.Subsequently,Fisher distance is used to extract features.Finally,classification of extracted features is performed by back-propagation neural networks. Four types motor imagery EEG of three subjects is classified respectively.The results show that the average classification accuracy achieves over 80%,and the highest is 88.1% on tongue movement imagery EEG.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第33期169-171,190,共4页
Computer Engineering and Applications
基金
江西省教育厅青年科学基金项目(No.GJJ09622)
关键词
身份识别
二阶盲辨识
运动想象
脑电
person identification
second-order blind identification
motor imagery
Electroencephalo gram(EEG)