基于MEMD的运动想象脑电信号的特征提取与分析

对于传统特征提取算法对运动想象脑电信号识别能力不足的问题,采用多元经验模式分解(multivariate empirical mode decomposition,MEMD)的方法用于分析运动想象的脑电信号.目前此方法主要应用在股票收益与宏观经济关系分析上,MEMD将标准经验模式拓展到多通道信号处理,适合于分析多元时间序列,并能够同时处理多通道的多尺度分解,进而在不同尺度下对多元时间序列的时间频率特性进行比较.通过Emotiv传感器对自定义的左右运动想象任务采集数据,采用MEMD提取相关脑电特征的边际谱,使用支持向量机对相关特征量进行分类.实验表明,此方法增强了定位脑电信号的频率信息的准确性,能够有效地提高对脑电信号的识别能力.

Feature extraction and analysis of imaginary movements in EEG based on MEMD

A method of multivariate empirical mode decomposition( MEMD) is proposed for low recognition rates when analyzing electroencephalogram( EEG) by using feature extraction algorithm of empirical mode decomposition( EMD) . Currently ,this method is mainly used in the analysis of stock market returns and macroeconomic . The MEMD is suitable for the analysis of multivariate time series ,which is expanding multichannel signal processing from the standard EMD . It extracts common modes from all channels in same-index intrinsic mode functions which allows the temporal-frequency features among different channels which can be compared in each subband. Through adopting MEMD to extract related marginal spectrum of EEG characteristics in the data of performing self-paced left and right motor imagery tasks by Emotiv sensor, support vector machine (SVM) classifier is used to classify the relevant features. The experiments show that the method enhances the locating accuracy of EEG frequency information; it could effectively improve the recognition ability of EEG signal.