基于EMD和Hilbert变换的自发脑电信号特征提取

EEG Feature Extraction Based on Empirical Mode Decomposition and Hilbert Transform in Brain Computer Interface

在脑机接口研究中,针对脑电信号的特征提取,提出一种基于EMD的Hilbert变换的方法。在变换过程中根据信号的局部特征自动选择基函数,求得信号在每个时间段的希尔波特谱;以时频窗口内的统计特性作为特征,利用Fisher距离选择最佳特征集输入分类器。最后利用BCI 2003竞赛数据,通过对特征矢量的可分性和识别精度两个指标的评估,表明了所提出方法的有效性。

In the study of brain computer interfaces,a method based on empirical mode decomposition （EMD） and Hilbert transformation was proposed. The method was used for the feature extraction of electroencephalogram. In this method, the basis function was selected automatically according to the local features of signal during the transforming process, the Hilbert spectrum was obtained in each period, and the statistical characteristics in time-frequency window were considered as features. Then the optimal feature sets were formed by the Fisher distance rule and put into the classification. The performance of the eigenvector was evaluated by separability and recognition accuracy with the data set of BCI 2003 competition,and classification results proved the effectiveness of the proposed method.