左右手想象运动的脑电信号分类识别

Classification and Identification of EEG based Imagery Left-right Hands Movement

利用脑电信号的事件相关同步/去同步(ERS/ERD)现象,针对脑电信号的非平稳随机特性,采用短时傅里叶变换(STFT)提取信号的时频特征,并分别用fisher分类器、人工神经网络(ANN)和支持向量机(SVM)对特征进行模式分类,正确率分别为66.2%(53/80),72.5%(58/80),81.2%(65/80)。实验结果表明STFT能有效提取脑电信号特征,且SVM是一种较优的分类方法。

Use brain-electrical signal＇s phenomena of event-related desynchmnization/event-related synchronization（ERS/ERD）,aim at its non-stationary random feature,adopt short-time Fourier transform to extract its time-frequently feature.After that,classify the feature with fisher,Artificial Neural Network and support vector machine,and respectively,The correct clssfication rates are 59 percent,70 percent,as well as over 80 percent.Experimental result manifest STFT there can be used for extracting characteristic of the brain electrical signal,and SVM is a great method relatively.