Electroencephalogram signal denoising method based on residual generative adversarial network

Abstract

The invention discloses an electroencephalogram signal denoising method based on a residual generative adversarial network. The electroencephalogram signal denoising method comprises the steps of selecting an electroencephalogram sample, constructing a noisy electroencephalogram signal sample, dividing a network training set and a network testing set, constructing the residual generative adversarial neural network, training the residual generative adversarial neural network and reconstructing a denoised electroencephalogram signal. Due to the fact that the residual error generation adversarial neural network is constructed, by introducing the residual error generator and the discriminator, the learning ability of the neural network is enhanced, real-time denoising is achieved, the discriminator is introduced, the electroencephalogram denoising efficiency and quality are improved, and effective features are screened out; a signal denoising process is divided into a model training process and a denoising process, and the signal-to-noise ratio and mean square error of signal denoising are improved. The method has the advantages of being simple in neural network structure, high in electroencephalogram signal denoising efficiency, good in denoising quality and the like, and can be applied to the preprocessing process of electroencephalogram signal processing and the technical field of signal denoising processing.

Description

technical field [0001] The invention belongs to the technical field of computers, and in particular relates to an EEG signal denoising method for residual generation against a neural network. [0002] technical background [0003] Brain-computer interface has a wide range of applications in education, health assistance, entertainment, and military affairs. However, the EEG signal needs to go through a series of processing steps such as signal processing from the beginning of acquisition to the process of being effectively applied. This is because the EEG signal is very weak, the unit of the EEG signal is microvolts, the anti-interference ability of the EEG signal is poor, and it is easily affected by various factors. In the process of collecting EEG signals, it is easy to be disturbed by the environment. Therefore, removing EEG interference signals, improving the signal-to-noise ratio of EEG signals, and extracting effective features in EEG signals are the prerequisites for e...

AI Technical Summary

Problems solved by technology

There are four different wavelet transform techniques in the EEG denoising field: discrete wavelet transform, dual-tree wavelet transform, double-density wavelet transform and double-density dual-tree wavelet transform and four different thresholding techniques (such as hard thresholding , soft threshold, semi-soft threshold and neighborhood coefficient threshold) to denoise the EEG signal damaged by the EMG signal. These methods filter out artifacts on the EEG signal and achieve good results, but these filters are excessive Depends on the adjustment of parameters, it is easily affected by parameters, and the artifact removal effect is unstable

Based on the above analysis, the existing methods cannot meet the EEG signal noise reduction requirements required for analysis

Images