约束独立分量分析及其在脑电信号伪差分离中的应用

Constrained ICA and Its Application to Removing Artifacts in EEG

独立分量分析(ICA)算法是一种运用统计方法,从一系列标准信号中提取独立成分的技术。由于脑电信号是由若干相对独立的成分组成,所以运用ICA算法来处理脑电信号受到广泛关注。本文介绍了一种新型的约束独立分量分析(cICA)算法,它能解决FastICA算法在源信号分离时输出排列无序性的问题。并通过实验表明,它在脑电伪差分离时可减少人工处理的影响,且具有良好的稳健性与较快的收敛速度。

Independent component analysis (ICA) is a statistic technique which extracts independent components from a set of standard signals. Since Electroencephalogram (EEG) signals are the mixture of several relatively independent sources, ICA has attracted extensive attention in the field of EEG processing. In this paper, a new Constrained ICA(cICA) algorithm is introduced, it would solve the problem of orderless output when FastICA algorithm is used. The experiment results testify that the cICA algorithm can redu...