眼电伪迹自动去除方法的研究与分析

发布时间：2018-03-06 04:00

本文选题：脑电接口　切入点：眼电伪迹去除　出处：《仪器仪表学报》2014年11期 　论文类型：期刊论文

【摘要】：脑电信号采集时很容易受到眼电信号的干扰,从而影响脑机接口系统的性能。为此,提出一种基于离散小波变换(DWT)和典型相关分析(CCA)的眼电伪迹自动去除方法,即DCCA法。首先,对采集的多导脑电信号和眼电信号进行离散小波变换,获得多尺度小波系数,并利用典型相关分析去除小波系数间的相关性,得到互不相关的典型小波系数;进而,利用相关系数判别眼迹成分,将相应典型小波系数置零并依次采用CCA逆变换和DWT逆变换重构剔除眼电伪迹后的脑电信号。基于9位实验者的4种眼电数据进行实验研究,并从统计学的角度对实验结果进行显著性检验。结果表明,DCCA法相对其他方法在均方根误差、信噪比方面具有显著优势,且具有较好的实时性,并表现出较强的适应能力。
[Abstract]:Eye-electric signals are easily interfered when EEG signals are collected, which affects the performance of BCI system. Therefore, an automatic Eye-electric artifact removal method based on discrete wavelet transform (DWT) and canonical correlation analysis (DCCA) is proposed. The multiscale wavelet coefficients are obtained by discrete wavelet transform on the collected multichannel EEG and eye signals, and the correlation between wavelet coefficients is removed by canonical correlation analysis, and the discrete typical wavelet coefficients are obtained. The correlation coefficient is used to distinguish the eye trace components, and the corresponding typical wavelet coefficients are set to zero and the EEG signals after eliminating the eye electrical artifacts are reconstructed by the CCA inverse transform and the DWT inverse transform in turn. The experimental study is carried out based on the four kinds of eye electrical data of the 9-bit experimenter. The results show that DCCA has significant advantages over other methods in root mean square error (RMS) and signal-to-noise ratio (SNR).
【作者单位】：北京工业大学电子信息与控制工程学院;
【基金】：国家自然科学基金(81471770,61201362) 北京市自然科学基金(7132021) 北京市教育委员会面上项目(KM201110005005)资助项目
【分类号】：TN911.4;TP334.7

【参考文献】