基于SVM和小波分析的脑电信号分类方法

EEG SIGNAL CLASSIFICATION METHOD BASED ON SVM AND WAVELET ANALYSIS

根据癫痫脑电信号与正常脑电信号波形和能量特征的不同,研究了两种的脑电信号分类方法,一种采用支持向量机SVM(Support Vector Machines)分类器对正常脑电和癫痫脑电进行分类;另一种使用小波分析和支持向量机相结合的方法对脑电进行分类,并比较了这两种方法对正常脑电和癫痫脑电分类的正确率。实验结果表明,小波分析和SVM结合的方法对脑电信号分类可以取得更好的效果,能有效区分癫痫脑电和正常脑电。

Due to the differences of EEG signal waveforms and spatial energy characteristics between epilepsy patients and healthy person,we employ two methods to classify these two kinds of signals.The one is to train and classify signals by SVM（Support Vector Machines） classifier,the other is the synthetic application of wavelet analysis and SVM.Then the classification accuracy of these two methods on normal EEG and epileptic EEG is compared.Results of the experiments show that the method of SVM combining wavelet analysis can achieve a better performance in EEG signal classification,which can effectively distinguish the epileptic EEG and the normal EEG.