基于Mahalanobis距离的运动意识分类研究  

Study of classification of motor imageries based on Mahalanobis distance

作　　者：李坤[1] 周晓兰[1] 郭晓静[1] 胡人君[1] 吴小培[1] 

机构地区：[1]安徽大学计算智能与信号处理重点实验室,安徽合肥230039

出　　处：《计算机工程与设计》2007年第7期1601-1603,共3页Computer Engineering and Design

基　　金：国家自然科学基金项目(60271024);安徽省人才基金项目(2004Z028);安徽大学模式识别创新团队基金项目。

摘　　要：提出基于Mahalanobis距离判别式算法的意识任务分类方法。对被测试者想象左右手运动时脑电信号的mu节律能量变化进行在线动态分析,提取EEG(C3,C4)两个通道的mu节律能量作为特征向量,用Mahalanobis距离判别式算法对左右手运动想象脑电模式进行分类,实验结果表明,正确识别率可达87.86%。The classification of metal activity based on Mahalanobis distance-based discriminant algorithm is proposed. The EEG signals are recorded during imagination of left and right hand movement.The mu rhythm energy of EEG and its online dynamic properties are analyzed. The mu rhythm of two channels （C3, C4） are extracted as feature vector,the event-related EEG patterns during left and right hand motor imagery are classified based on Mahalanobis distance-based discriminant algorithm. According to the analysis and experiment results, the correct rate of classification achieve 87.86 %.

关 键 词：马氏距离 脑电 脑机接口 特征提取 分类 

分 类 号：TP39[自动化与计算机技术—计算机应用技术] R318[自动化与计算机技术—计算机科学与技术]