The Spontaneous EEG Feature Extraction Based On WICA Method

With today’s social development and people’s living standards increasing, more and morepeople have their own vehicle, thus a lot of people lose or even completely lose athletic ability.In order to facilitate the daily life of these patients, a growing number of laboratories began tostudy on brain-computer interface (BCI). As a new interactive system, BCI collects EEGsignal through EEG acquisition device and extracts EEG feature for classification, andcombine different thinking activities with different instructions to achieve the contectbetween the human brain and peripheral. The appearance of the BCI, making the idea of usingmind to control peripherals directly as possible.EEG feature extraction is the most important step. How make people’s ideas expressed inthe form of electricity signal and control peripherals become a research hotspot. EEG whoseessence is the spontaneous EEG was thinking person performing certain mental activitiesarising from the EEG signal. For EEG signal acquisition, the electrodes placed on the scalpsurface not only records the electrical activity of neurons in the brain, but also tainted variousinterfering signals (artifact), such as frequency interference, blink, eye movement, ECGinterference and EMG interference. Therefore, EEG noise cancellation is a key issue in theEEG signal processing.How to extract useful information from EEG is a focus and difficult processing. Themodern method of wavelet transform and independent component analysis method proposedprovides a new way for the analysis of EEG. To extract the characteristic wave imprecise dueto the effects of noise, using wavelet packet to extract energy feature is not obvious.Independent component analysis is based on multivariate statistical properties of the signalanalysis and processing, multi-channel mixed-signal can be independent separation.Extracted by taking into account the characteristics of wave is an independent component ofthe mixed-EEG, Independent component analysis for the characteristic wave can be isolatedto a certain extent. However, due to the uncertainty of ICA, there are some limitations.This article is based on a single task for spontaneous motor imagery EEG featureextraction. We propose a new EEG feature extraction method based on infor-fast ICA algorithm combined with the wavelet transform (Wavelet Transform) method named WICAmethod. Through the ICA algorithm EEG preprocessing, wavelet transform for the first timefeature extraction, and then use infor-fast ICA algorithm for secondary feature extraction, andultimately extracted the EEG mu rhythm.Comparing the result of EEG feature extracting used the method in this article with theICA method, wavelet transform method and traditional ICA and wavelet transform method. Itis showed that the new method in this article to extract single-task motor imagery EEG in10-12HZ whose energy feature is most obvious. Final results verify the effectiveness of thenew method we used.