EMD和CSP融合的情绪脑电特征提取方法  

Feature Extraction of Emotional EEG Based on Fusion of EMD and CSP

作　　者：王玫 郑威[1] 杨双竹 WANG Mei;ZHENG Wei;YANG Shuangzhu(School of Electronic Information,Jiangsu University of Science and Technology,Zhenjiang 212003)

机构地区：[1]江苏科技大学电子信息学院,镇江212003

出　　处：《计算机与数字工程》2024年第7期2082-2088,2114,共8页Computer & Digital Engineering

摘　　要：脑电(EEG)是一种非线性、非平稳的时变信号。为提取有效的情绪脑电特征,论文提出了经验模态分解(EMD)和共空间模式算法(CSP)融合的方法,在公开数据集DEAP和SEED进行积极情绪和消极情绪的二分类实验。首先通过实验选取最优CSP特征参数;其次将IMFs作为CSP的输入,提取具有时-频-空三个维度特性的情绪特征,通过支持向量机(SVM)进行验证;最后进一步分析了不同脑区与情绪之间的关系。实验结果表明,m=5和m=4分别是DEAP和SEED的最优CSP特征参数;且低频的本征模态函数(IMF)分量更易区分消极和积极情绪,EMD-CSP方法在DEAP和SEED数据集上分类准确率分别达到了88.7%和99.2%。EEG is a kind of non-linear and non-stationary time-varying signal.In order to extract effective features of emo⁃tional EEG,this paper proposes a fusion method of empirical mode decomposition(EMD)and common spatial pattern algorithm(CSP),and conducts two classification experiments of positive and negative emotions on open data sets deap and seed.Firstly,the optimal CSP feature parameters are selected through experiments.Secondly,IMFs are used as the input of CSP to extract emotion features with three dimensions of time frequency space,which are verified by support vector machine(SVM).Finally,the relation⁃ship between different brain regions and emotion is further analyzed.The experimental results show that m=5 and m=4 are the opti⁃mal CSP characteristic parameters of deap and seed respectively.And the low-frequency intrinsic mode function(IMF)component is easier to distinguish negative and positive emotions.The classification accuracy of EMD-CSP method on deap and seed data sets reaches 88.7%and 99.2%respectively.

关 键 词：脑电信号 情绪识别 经验模态分解 共空间模式算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]