基于能量法的脑机接口运动想象分类研究

Research of Classification about BCI Based on the Signals Energy

针对脑机接口运动想象脑电信号的分类识别问题,提出了一种基于小波包分解的C3、C4二通道能量特征提取方法。该方法首先采用6阶的巴特沃斯带通滤波对二通道脑电信号进行降噪;然后采用Daubechies类小波函数对其进行5层分解,选择第四层CD4、第五层CD5的小波系数进行重构并抽取其能量特征;最后采用线性距离判别进行分类和使用Kappa系数进行分类衡量。利用BCI2008竞赛的标准数据BCICIV_2b_gdf进行验证,结果表明利用该方法可以较好地反映事件相关同步和事件相关去同步的特征,为BCI研究中事件相关电位的分类识别提供了有效的手段。

Aiming at the issue of motor imagery electroencephalography（EEG） pattern recognition in the research of brain-computer interface（BCI）, a power feature method based on discrete wavelet packet decomposition is proposed for the channels C3 and C4. Firstly, a six-border Butterworth filter is used to denoise the two-channel EEG signals. Secondly, two-channel EEG signals are decomposed to five levels using Daubechies wavelet and the fourth level and the fifth level are chosen to reconstruct the signals and compute its power feature. Final y, linear discriminant analysis （LDA） is utilized to classify the feature and the Kappa value is utilized to measure the accuracy of the classifier. This method is applied to the standard dataset BCICIV_2b_gdf of BCI Competition 2008, and experimental results show that this method reflect the feature of event-related sychronization and event-related desychronization obviously and it is an effective way to classify the EEG patterns in the research of BCI.