A Method for Extracting Independent Components of EEG Signals Based on Convolutional Blind Source Separation

Abstract

The invention discloses a method for extracting independent components of EEG signals based on convolution blind source separation. The specific steps are: building an AD sampling module, a short-time Fourier transform module, a frequency domain instantaneous blind source separation module, and sequence adjustment EEG signal independent component extraction system based on convolutional blind source separation of module and short-time Fourier inverse transform module; AD sampling module samples EEG signal; short-time Fourier transform module transforms time-domain EEG signal to the frequency domain; the frequency domain instantaneous blind source separation module separates the frequency domain instantaneous mixed signal; the sequence adjustment module sequentially adjusts the independent components in the vector on each frequency domain segment; the short-time Fourier inverse transform module converts the frequency domain The separation results are transformed into independent components in the time domain. The invention extracts independent components of EEG signals based on a more realistic convolution mixed model, adopts a frequency domain algorithm of convolution blind source separation, and has simple implementation, good separation effect and low computational complexity.

Description

technical field [0001] The invention relates to a method for extracting independent components of electroencephalogram signals, in particular to a method for extracting independent components of electroencephalogram signals based on convolutional blind source separation. Background technique [0002] EEG signal feature extraction is one of the key steps in brain-computer interface. Separating independent components from multi-channel EEG signals, and then extracting feature signals from independent components is a widely used EEG feature extraction process. The EEG signal independent component extraction method models the multi-channel EEG mixing process, and performs blind source separation on the EEG mixed signal based on the model, and obtains multiple components that are statistically independent. At present, the extraction method of independent components of EEG signals is generally based on the instantaneous mixing model, that is, it is assumed that multi-channel EEG ...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a method for extracting independent components of EEG signals based on convolutional blind source separation, so as to solve the problem that the independent component extraction method based on the instantaneous mixing model cannot truly describe the mixing process of EEG signals, thereby affecting the feature extraction and classification of EEG signals The problem of recognition accuracy