基于DTCWT的运动想象脑电特征提取  

MOTOR IMAGINATION EEG FEATURE EXTRACTION METHOD BASED ON DTCWT

作　　者：汤伟[1,2] 耿逸飞 Tang Wei;Geng Yifei(College of Electrical and Control Engineering,Shaanxi University of Science and Technology,Xi’an 710021,Shaanxi,China;Institute of Industrial Automation,Shaanxi University of Science and Technology,Xi’an 710021,Shaanxi,China)

机构地区：[1]陕西科技大学电气与控制工程学院,陕西西安710021 [2]陕西科技大学工业自动化研究所,陕西西安710021

出　　处：《计算机应用与软件》2023年第4期80-84,106,共6页Computer Applications and Software

基　　金：陕西省重点科技创新团队计划项目(2014KCT-15);陕西省科技统筹创新工程计划项目(2012KTCQ01-19)。

摘　　要：针对脑电信号采用单一特征识别存在自适应性差和识别率低等问题,提出一种基于双树复小波(DTCWT)的多特征融合的左右手运动想象脑电特征提取方法。对原始脑电信号进行DTCWT变换提取最佳时频段;对所提取的信号频段进行希尔伯特变换与Lempel-Ziv复杂度计算,将得到的时-频域特征与非线性特征组合为特征向量;采用线性判别分析(LDA)完成运动想象任务的分类。实验采用BCI CompetitionⅢ竞赛数据对该方法进行验证,仿真结果表明其识别准确率明显提高,最高可达89.84%。Aiming at the problems of poor adaptability and low recognition rate in single feature recognition of EEG signals,this paper proposes a method for extracting EEG features of left and right hand motor imagination based on dual-tree complex wavelet(DTCWT)multi-feature fusion.The DTCWT transform was performed on the original EEG signal to extract the best time frequency band.The Hilbert transform and Lempel-Ziv complexity calculation was performed on the extracted signal frequency band,and the time-frequency domain characteristics and nonlinearity characteristics calculated respectively were combined as feature vector.The linear discriminant analysis(LDA)was used to complete the classification of the motor imagination task.The experiment used BCI CompetitionⅢdata to verify the proposed method.The simulation results show that its recognition accuracy is significantly improved,up to 89.84%.

关 键 词：脑电信号 运动想象 双树复小波变换 Lempel-Ziv复杂度 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]