The Study Of Removal Of Ocular Artifacts And Selection Of Reference Electrodes In Eeg Signals

The EEG signals are the spontaneous and rhythmic electrical activity of the brain cell,we usually refer to is the scalp EEG.As a non-invasive detection method,the research scope covers neuropsychology,pathology,cognitive neuroscience,social psychology,signal and information processing and other fields.However,the EEG signal is highly random,and has many types of rhythms and high sensitivity in times,therefore it can easily be contaminated by noise and produce various artifacts,which brings lots of errors to follow-up analysis and research.Therefore,how to pretreating is very important for theoretical and practical.In this paper,a detailed study was made,that are the removal of eye artifacts and the selection of reference electrodes during the pretreatment.The eye electric signal is the basis of the traditional method of removing the traditional eye power artifacts,which is bound to adulterate other noise in the acquisition process and need to be identified manually.In order to achieve automatic recognition and removal of the eye electrical artifacts,an automatic detection method based on FastICA is proposed.This method extracts the independent components of the signal through the FastICA method,calculates the GFP(global field power)value of the signal,and then calculates the correlation coefficient between the independent components and the GFP values.By comparison,the independent component corresponding to the correlation coefficient of the maximum absolute value is the independent component of the eye power artifacts.Finally,the independent component is reset to reconstruct the clean brain signal to realize the automatic removal of the eye power artifacts.Based on this,we propose a method to eliminate the overestimation problem,the method prevents the loss of information in the EEG signal effectively and retains the useful information.The results show that the method can automatically identify and remove the eye power artifacts on the premise of retaining useful EEG information.The reference electrode standardization technology(REST)has been widely applied to the field of brain signal acquisition and analysis since its introduction.Through the method,the potential value that based on a point or the average reference electrode can be converted into a potential value that is referenced to a spatially infinity point,that is the zero-reference electrode.In this paper,based on the wavelet packet method,we selected the energy of node,node energy,wavelet packet entropy and the energy of ?,?,?,? as the feature quantities,and compared the classification accuracy of the zero-reference electrode and the average reference electrode method on two sets of data.We confirmed the reliability of the reference electrode standardization technique in practical applications.