【摘要】 提出了一种基于ICA和局部能量最大的单次自动提取脑电信号中N2和P3成分的方法。实验采集了七名健康受试者在观察"模拟阅读"刺激界面状态下的32导的脑电信号,利用Fast ICA算法对单试次脑电信号进行盲源分离,将得到的32个分量用样本方差最大方法,在固定时间段自动提取脑电信号中N2和P3成分。把N2、P3分量直接作为单次提取的特征,利用支持向量机进行分类,同时和最优单通道时域特征的分类进行对比。结果表明:基于ICA方法可以有效地自动提取单次脑电信号中N2和P3成分,且分类效果比最优单通道有显著提高。更多还原

【Abstract】 This paper introduces a method of single automatically extract EEG of N2 and P3 component based on ICA and big-gest local energy. Experiment collectes seven healthy people 32 lead EEG in the state of observed at"Imitating-Reading"stimula-tion interface. Firstly,single trial EEG is taken into blind source separation using the Fast ICA algorithm to 32 component result. Sec-ondly,automatic extraction N2 and P3 components in the 32 component result at the fixed time period,using maximum sample vari-ance method. The N2,P3 component directly are as a single extraction feature and SVM classification method is used to classify,atthe same time compared with the classification that using characteristics parameters of best single-channel in the time-domain. Theresults indicate that using automatic extraction of a single EEG N2 and P3 component based on ICA method is effective and classification results has risen considerably than the best single channel.更多还原