运动想象脑电信号特征的提取与分类

Features Extraction and Classification of Motor Imagery EEG

为了能更好地对左右手运动脑电信号进行检测,用VC设计了脑电处理的上位机系统,利用小波变换和SVM的方法对脑电信号进行研究。这种方法主要是通过小波分解,针对C3电极处脑电信号,将脑电信号中各成分分别突出到不同尺度。提取不同尺度上的脑电信号的两种特征,即变化系数和波动指数。然后再将这两种特征组成一个向量送到支持向量机中进行训练和检测。共选取了四种采样频率,准确率最高达到98.32%。

This paper uses VC designed the EEG processing system,then used the method of wavelet transform and SVM to study EEGin order to better research movement imagination EEG signals.This method is mainly through the wavelet decomposition,the components of the EEG are displayed in different scales at the C3 electrode.Extracted two feature of the EEG signal on different scales,that iscoefficient of variation andvolatility index.The two kinds of feature formeda vectorand then move tothe SVM to training and testing.Selects four kinds of sampling frequency,the highest accuracy rate reached 98.32%.