基于小波包分解脑电信号特征提取.pdfVIP

第 28卷 第 12期 仪 器 仪 表 学 报 Vol1 28 No1 12 2007年 12月 Ch inese Journal ofScientific Instrum ent Dec. 2007 , , (上海交通大学电信学院仪器系 上海 200240) : 在脑机接口研究中, 针 脑电信号的特征抽取, 提出一种基于小波包分解的方法, 利用 F isher距离准则, 选择具有较大 可分离性的特定子带小波包系数和能量作为有效特征, 构成特征矢量, 并采用 BCI 2003竞赛数据, 通过 该特征矢量的可分性 和识别精度 2个指标的评估, 表明了所提出方法的有效性。 : 脑机接口; 小波包分解; 特征提取; 子带能量 : R318 TP391. 4 : A : 460. 4025 510. 8050 EEG feature extraction in brain computer interface based on wavelet packet decomposition Wu Ting, Yan Guozheng, YangBanghua ( School of E lectron ic, Inform a tion a nd E lectrica l E ng. , Sha ngh a i J iao Tong Un iversity, Sha ngh a i 200030, Ch ina ) Abstr act: In the stu y of brain computer interface, a metho base on wavelet packet ecomposition is propose which is use for the feature extraction of electroencephalogram. The power of special sub2ban s an coefficients of wavelet packet ecomposition that have larger separability are selecte to construct eigenvector accor ing to Fisher istance criterion. The eigenvector is obtaine by comb in ing the effective features of electroencephalograph signals from ifferent channels. The performance of the eigenvector is evaluate by separability an recogn ition accuracy u2 sing the ata set from BCI 2003 competition. Classification results have prove the effectiveness of thismetho . K ey words: brain computer in terface (BCI); wavelet packet ecomposition (WPD); feature extraction; power of sub2ban EEG具有信息量大、时间分辨率高、设备便于携带、价格 1 低廉以及无创性等优点, 为脑机接口提供了实际可行性,