基于脑电信号的身份识别方法  

Person identification based on electroencephalogram signals

作　　者：胡剑锋[1] 包学才[1] 

机构地区：[1]江西蓝天学院信息技术研究所,江西省南昌市330098

出　　处：《中国组织工程研究与临床康复》2009年第17期3260-3264,共5页Journal of Clinical Rehabilitative Tissue Engineering Research

基　　金：江西省自然科学基金项目(2008GQS0003)~~

摘　　要：基于脑电信号的身份识别是通过采集试验者的脑部信号来进行身份认证。对于同一个外部刺激或者主体在思考同一个事件的时候,不同人的大脑所产生的认知脑电信号不同。选取与运动意识想象有关的电极后,分析不同个体在特定状况下脑电的个体差异,采用以回归系数、能量谱密度、相同步、线性复杂度多种信号处理结合方法对运动想象脑电信号进行处理来进行特征提取。组合多元特征向量并运用多层BP神经网络对不同个体的脑电信号进行分类,并在不同的意识想象及不同数据长度、不同的波段对试验者进行识别率验证分析。结果表明,不同运动想象的平均识别率均在80%以上,其中以想象舌头运动的识别率较高,达到90.6%,不同波段的识别率也表明意识想象的模式及相应波段对身份认别有较大的影响。Person identification based on electroencephalogram （EEG） signals is to identify a person using their EEG signals. EEG signals of different people have individual differences to the same stimulation or same motor imagery. After selecting electrodes corresponding to motor imagery, individual differences of EEG of different people under specific condition were analyzed, and several signal processing methods were adopted for feature extraction, such as regression coefficient, phase synchronization, energy spectral density and linear complexity. Muitilayer BP neural network was used for classification of person identification by combining multi-feature vector. IdentJficatJon effect was analyzed for different data length and wave band. Results show that mean identification rate was more than 80% for different motor imagery, in particular identification rate of tongue movement, over 90.6%, and that identification rate depends on the paradigm of motor imagery and wave band.

关 键 词：脑电信号 身份识别 运动想象 回归系数 

分 类 号：R319.1[医药卫生—基础医学]