Study On Multi-channel EEG Signal Denoising Algorithm

Electroencephalogram(EEG)is a popular diagnostic tool for diseases.It is often used to monitor the changes of EEG signals to help people better understand the physiological structure of the brain.Medical workers can analyze and diagnose EEG according to their experience,but EEG signals contain a lot of noise,so the removal of noise is the primary link of EEG data analysis and processing.How to recover the original signal from the noisy signal and realize the separation of signal and noise has become an important research topic of EEG denoising.Wavelet analysis is a new technology developed rapidly in the field of multi-channel EEG signal denoising in recent years.It is widely used in clinical diagnosis and scientific research.Many scholars also use wavelet function to study EEG denoising.However,the current denoising algorithm has some limitations in noise recognition and noise removal.In view of this,based on the wavelet denoising algorithm,this paper proposes a new denoising method of wavelet threshold function.At the same time,Bayes is used to estimate the coefficients of the new wavelet threshold function,and the denoised EEG signal is analyzed by Kalman filter.The main research work is as follows:Based on the characteristics of multichannel EEG signal,PCA dimension reduction technology is used to eliminate the information redundancy between multichannel EEG signals.The denoising algorithm and influence of Fourier,wavelet decomposition and reconstruction,wavelet soft threshold are discussed.A new threshold wavelet function is constructed to solve the problem of unsatisfactory denoising effect of EEG signal under the condition of multiscale decomposition.Through the simulation experiment,the denoising effect of the new threshold wavelet function algorithm is verified,and the influence of different algorithms on the denoising performance is compared from the signal evaluation index.Based on the new wavelet threshold denoising algorithm,the Bayesian estimation theory is analyzed according to the Laplace noise model of multichannel EEG signal.Starting from the Bayesian estimation algorithm,it is proved that the algorithm can effectively estimate the wavelet coefficients of the new threshold function.The experimental results show that compared with the new threshold wavelet function de-noising algorithm,the Bayesian algorithm can effectively solve the problem of low energy.From the analysis of signal evaluation index,the energy ratio is increased by 12%,the root mean square error is reduced by 0.1,the signal-to-noise ratio is increased by 0.2db,and the peak value of spectrum is closer to the original signal.The EEG signal denoised by Bayesian estimation algorithm is tested,and the state equation of the noise signal is evaluated to select the optimal denoising solution for Kalman filtering.The experimental results show that compared with Bayesian estimation algorithm,Kalman filtering algorithm can effectively filter out the noise with amplitude less than 5mv,and the signal-to-noise ratio is improved by 0.3d B,the energy ratio is close to 100%,which can better restore the original signal characteristics.