基于模糊熵的运动想像脑电信号特征提取  

Motor imagery EEG feature extraction based on fuzzy entropy

作　　者：田京[1] 罗志增[1] 

机构地区：[1]杭州电子科技大学机器人研究所,浙江杭州310018

出　　处：《华中科技大学学报（自然科学版）》2013年第S1期92-94,98,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61172134)

摘　　要：提出了一种基于模糊熵的运动想像脑电特征提取的新方法.根据脑电信号模糊熵时间序列良好的事件相关同步/事件相关去同步(ERS/ERD)特征,构筑人脑想像左右手运动任务时的C3和C4脑电模糊熵差值特征向量,利用Fisher线性分类器对左右手运动想像的意识任务进行分类.对国际脑机接口竞赛相关数据进行了测试,最大分类准确率达到87.22%.基于模糊熵的脑电信号特征可以作为有效特征对运动想像任务进行区分.Fuzzy entropy has been applied to the classification of motor imagery.The fuzzy entropy features was able to indicate the event-related synchmnization/event-related desynchronization(ERS/ ERD)time sequence changes effectively,and the difference between C3and C4EEG(encephalon electrical signal)fuzzy entropy was used as the feature of the left/right hand motor imagery tasks.Then, a linear fisher criterion classifier method was applied to classify the left/right hand motor imagery tasks.The results of simulation on the data of the 2003BCI contest show that the best accuracy is 87. 22%.The classification of consciousness tasks will be more effective by using the feature based on fuzzy entropy of EEG.

关 键 词：脑电信号 模糊熵 运动想像 特征提取 事件相关同步/事件相关去同步 

分 类 号：TN911.7[电子电信—通信与信息系统]