单通道脑电信号眼电伪迹去除算法研究

由眨眼和眼动产生的眼电伪迹（Electrooculography，EOG）信号是脑电信号（Electroencephalography，EEG）中的主要噪声信号之一.目前，多通道脑电信号中眼电伪迹的去除算法已经较为成熟.而在单通道脑电信号的眼电伪迹去除中，由于采集通道数量较少且缺乏参考眼电信号，目前尚无十分有效的去除方法.本文提出一种基于小波变换（Wavelet transform，WT）、集合经验模态分解（Ensemble empirical mode decomposition，EEMD）和独立成分分析（Independent component analysis，ICA）的WT-EEMD-ICA单通道脑电信号眼电伪迹去除算法.实验表明:WT-EEMD-ICA算法有效地解决了单通道WT-ICA算法中的超完备问题，能够有效去除单通道脑电信号中的眼电伪迹，并且分离出的眼电伪迹成分与参考通道采集的眼电信号相关性较强.

EOG Artifact Removing Method for Single-channel EEG Signal

Electrooculography (EOG) artifacts generated by eye movements and blinks are the major artifacts in electroencephalography (EEG). There are many common effective methods for removing the multi-channel EEG artifacts. However, due to the limitation of input channel number and the lack of reference EOG signal, there is no very effective artifact removing method for single-channel EEG signal. In the present study, a novel EOG artifact removing algorithm WT-EEMD-ICA for single-channel EEG signal is proposed based on wavelet transform (WT), ensemble empirical mode decomposition (EEMD) and independent component analysis (ICA) technologies. The result shows that the WT-EEMDICA method, which successfully solves the overcomplete problem of WT-ICA in single channel artifact removal, can separate the EOG and EEG successfully only from one single-channel EEG, and that the useful information involved in original EEG signal can be greatly saved.