摘要
为在脑机接口系统BCI(brain-computerinterface)中有效选择导联进行特征提取和分类提供依据,研究了基于运动想象脑电信号的导联排序。根据公共空间模式算法CSP(common spatial pattern)原理提出了一种导联排序方法——基于协方差和主成分分析的排序算法CPSorting(covariance and principal component sorting),并研究了运动想象脑电信号MI(motor imagery)中导联的排序情况以及排序靠前的导联对分类的贡献。利用公共空间模式算法对CPSorting排序后导联的数据提取特征,再分别应用支持向量机SVM和K近邻算法KNN进行分类。实验结果表明了该排序算法能有效地对基于运动想象脑电信号的导联进行排序。
In order to provide the basis for selecting channels effectively to do feature extraction and classification in brain-computer interface system, a channel sorting technique called CPSorting algorithm based on covariance and principal component analysis is presented according to the theories of common spatial pattern algorithm (CSP). At the same time, the sorting situation of motor imagery (MI) EEG and the contribution to the classification of channels on the top of the list are discussed. The data received from the channels after sorting are used for extracting features with CSP algorithm and classifying with SVM and KNN algorithms. The experimental results prove that the presented algorithm can effectively sort the channels ofMI EEG.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第19期4265-4267,4292,共4页
Computer Engineering and Design
基金
上海市科委重点科技攻关基金项目(08511501702)
上海市重点学科建设基金项目(J50103)