基于两类运动想象及SVM脑电识别的小波特征窗选研究
首发时间:2018-06-12
摘要:本文研究了在使用SVM方法进行脑电识别的过程中信号特征的加窗选取问题,以更多地获取分类性质显著的脑电特征,同时提高脑电识别效果,其中信号特征从信号的小波变换中选取。文章以两类运动想象脑电信号为研究对象,通过对比窗选特征向量与单值特征对应的识别效果来分析窗选特征向量的分类性质。研究表明窗选特征向量对应的识别效果普遍好于单值特征,同时给出了如何选取大量特征组成特征向量的方法,研究还分析了特征向量对应的识别效果受窗选方式的影响,研究体现了小波变换的窗选特征向量在脑电识别中的实用价值。
For information in English, please click here
Research on Windowing selection to Wavelet Transform Features Based on Two-Type Motor Imagery EEG and SVM Identification
Abstract:The paper studied windowing selection of EEG features in the process of EEG identification by SVM, in order to obtain more EEG features with notable classifying property, and improve the effect of EEG identification at the same time. EEG features were selected from wavelet transform of EEG. In the paper, two-type motor imagery EEG was used. By contrasting between the identification effects of windowing-select feature vector and single feature, classifying properties of feature vector were analyzed. Study showed that the identification effect of windowing-select feature vector was generally better than that of single feature, and also provided the method of how to select a large number of EEG features to constitute vector. Study also analyzed the influence to identification effects of feature vector by windowing selection styles. Study displayed practical values of windowing-select feature vectors of wavelet transform in EEG identification.
Keywords: BEI (Biotic Electric Interface) EEG (electroencephalogram) SVM (support vector machine) wavelet transform CR (conformity rate)
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于两类运动想象及SVM脑电识别的小波特征窗选研究
评论
全部评论0/1000