基于PCA及SVM的运动想象脑电信号识别研究

为了解决脑机接口(BCI)中不同意识任务下运动想象脑电信号的分类问题,提出了一种基于PCA及SVM的识别方法。针对Hilbert-Huang变换和AR模型提取的脑电信号特征,首先采用主成分分析PCA对高维特征向量进行降维处理,然后用支持向量机进行分类。最后将本方法分类结果和Fisher线性分类、概率神经网络分类结果进行比较。实验结果表明,该方法分类正确率较高,复杂度低,具有一定的有效性,可用于脑机接口中。

Classification of Motor Imagery EEG Based on PCA and SVM

In order to solve the problem of the electroencephalogram （EEG） classification under different imagery task in brain computer interfaces （BCI）, a new recognition method based on principle component analysis （PCA） and support vector machine （SVM） is presented in this paper. Four features of motor imagery EEG signals extracted by combining the HHT with AR model, first, PCA was utilized to reduce dimensions of the high dimensional feature vectors. Then, SVM was used to classify different EEG patterns of motor imagery. Finally, this method was compared with Fisher LDA （linear discriminant analysis ） and probabilistic neural network （PNN）. Experimental results showed that the proposed method could classify different EEG patterns of motor imagery effectively due to its higher classification accuracy and lower complexity so as to be utilized in online BCI system.