基于HHT和SVM的运动想象脑电识别

对运动想象脑电信号(EEG)分类识别是脑-机接口(BCI)研究领域的重要问题。本文通过经验模式分解(EMD)将EEG分解为一系列内蕴模式函数(IMF),并对重要IMF的瞬时幅度提取AR模型参数,同时对所有的IMF进行Hilbert变换(HT)得到Hilbert谱,进而求得瞬时能量(IE)。将得到的AR参数和IE,结合时域均值和中值绝对偏差估计(MAD),组成初始特征,然后利用经遗传算法(GA)优化的支持向量机(SVM)进行分类,得到识别结果。对2008年BCI CompetitionⅣDataset 1中想象左手和脚运动的两组数据进行识别,在仅仅使用少数通道的情况下,识别正确率分别达到84.7%和85.8%,初步验证了该方法的有效性。

Discrimination of movement imagery EEG based on HHT and SVM

Discrimination of movement imagery event-related EEG is an important issue in brain-computer interfaces(BCIs).In this paper,empirical mode decomposition(EMD)is employed to decompose the EEG into a series of intrinsic mode funetions(IMFs),and then AR model parameters of instantaneous amplitude of some important IMFs can be obtained.Hilbert transform(HT)is conducted on all of the IMFs to get Hilbert spectrum,from which instantaneous energy(IE)can be gained.The obtained AR parameters and IE,combined with the mean of signal magnitude and median absolute deviation(MAD)estimate,form the initial features.Then support vector machine(SVM)optimized by genetic algorithm(GA)is adopted to achieve the optimal classification result.Two data sets on left hand and foot imagery of BCI Competition Ⅳ 2008 Dataset 1 are selected to carry out discrimination,and the discrimination results are 84.7% and 85.8%,respectively in the case of just using a small number of channels,which indicate the efficiency of this algorithm.