一种快速的脑电信号特征提取与分类方法

Fast Method for EEG Feather Extraction and Pattern Classification

在脑机接口研究中,针对两种思维任务的特征抽取和分类,提出一种以通道信号在两种类别下均值之间的欧式距离为特征,用阈值判定进行任务分类的方法;采用BCI2003竞赛数据,分析了该方法的实验背景和理论依据,并将分类精度与竞赛的最好结果进行了比较;表明了所提出方法运用在实际系统中的有效性和优越性。

In the study of brain-computer interfaces, a method of feature extraction and classification used for two kinds of imaginations was proposed. It considered Euclidean distance between mean traces recorded from channels of the two kinds as a feature, and determined imagination classes using threshold value. The background of experiment and theoretical foundation were analyzed referring to data sets of BCI 2003, and the classification error was compared with the best result of the competition. It is showed that the method has a better effectivity and advantage for applying to practical systems.