A Method of Denoising Multi-channel EEG Signals Fast Based on PCA and DEBSS Algorithm

Abstract

A method of de-noising multi-channel EEG signals which combines the principle component analysis (PCA) with density estimation blind source separation (DEBSS) is proposed in this paper. Based on removing high frequency noise in wavelet analysis, the PCA algorithm is used to process the EEG signals to reduce the data dimension. Then, the DEBSS algorithm is adapted to separate the EEG signals which data dimension has been reduced. The main interference is identified and removed by using cross-correlation coefficient and related non-linear parameters to analyze the independent components. Finally, through reconstructing the remaining independent components, the EEG signals without main interference will be obtained. The experimental results show that this method can eliminate the main interference of multi-channel EEG signals rapidly and effectively, meanwhile, it is stable and has strong scalability.