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。对三位受试者的四类运动想象脑电信号分别进行了分类识别,结果显示,四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。
Subject identification can be done by classifying motor imagery EEG signals.To achieve the identification over the subjects,firstly we adopt a blind source separation（BSS） method called second-order blind identification to process the motor imagery EEG signals and increase the signal-to-noise ratio of those signals.Subsequently,Fisher distance approach is used to extract features from the treated signals.Finally,classification of extracted features is performed by back-propagation neural networks.Classified identification results from the four types of motor imagery EEG of the three subjects show that the average classification accuracy reaches 80%,and the most accurate is tongue movement imagery EEG,which reaches 88.1%--a high level compared with that of similar studies.
出处
《江西蓝天学院学报》
2010年第2期45-49,共5页
Journal of Jiangxi Blue Sky University
基金
江西省教育厅青年科学基金项目(编号:GJJ09622)
关键词
身份识别
二阶盲辨识
运动想象
脑电
identification
second-order blind identification
motor imagery
EEG
