Emotion Recognition Based On Eeg Signal Late Positive Potential

How to accurately interpret human emotional state is a problem in the field of artificial intelligence,medical,computer and other workers.In the field of emotion recognition research,researchers have made some progress by continuous efforts,but we also need to try.At present,the emotion recognition based on EEG research heat is growing.It makes EEG signals collection more convenient and quick that Non-invasive electrode was applied.The current people's work and life pressure are larger,there are a lot of potential psychological problems.Emotion recognition based on EEG research could identify the individual emotion types of EEG signal,although the individual mood could be camouflaged on the surface,EEG signals can reflect the individual's state of mind to help the next adjustment and solve the potential problems for their own physical health and mental health.At the same time,Recognition of emotions provides a way for the brain electrical signal for human-computer interaction of emotional intelligence and disease treatment.Emotion recognition researches of EEG signals had a period of time,the main research method concentrated in the time domain,frequency domain and time-frequency domain,statistics analysis and discrete and continuous properties classification model,etc.,few people focus on Emotional nature of the EEG characteristics.EEG Signal late positive potential(300 ~ 1500 ms)occurs a few hundred milliseconds after the stimulus presentation for a few seconds,ERP late positive potential are involved in emotional processing,it has the biggest effect on brain area location of middle of the central-top area,the ERP late positive potential is an important emotional recognition characteristic.This paper will study EEG Signal late positive potential overlooked,explore to identify emotion effect of the EEG Signal late positive potential feature recognition.In this paper,the main work is as follows:First of all,Collect EEG signals and preprocess EEG signals before feature recognition and analysis.In this part mainly includes the design of the experiment of EEG acquisition,selection of subjects and EEG signal acquisition.Experiment according to the research content of this paper design to induce subjects to produce positive,neutral and negative emotions by the picture material.After collecting EEG signals,EEG signals are preprocessed by mainly using independent component analysis method to remove all kinds of artifact and noise,and keep maximum EEG components.Secondly,extract EEG signal feature and analyze.We extract EEG Signal late positive potential feature in this paper,and extract the three time-frequency characteristics to analyze.Through to the average event related potential variance analysis,it revealed that LPP_early(300 to 600 ms)LPP_medium(600 to 1000 ms)and LPP_late(1000 to 1500 ms)had significant difference on some leads.LPP_early(300 to 600ms)of three kind of emotions characteristics are significantly different in the occipital lobe,temporal lobe and parietal lobe,it shows that late positive potential are participated in the subjects of emotional response and adjustment,late positive potential characteristics can classify positive? neutral? negative emotions.Finally,classify and analyze emotion of EEG signals.We classify two types of emotional characteristics test by using support vector machine(SVM)algorithm and K nearest neighbor classification algorithm.LPP_early(300 ~ 600 ms),LPP_medium-(600 ~ 1000 ms)and LPP_late(1000 ~ 1500 ms)features can identify the positive,neutral,and negative emotion types in EEG signals,LPP_late identification accuracy is highest in the beta band,reached 83.33%,it showed that late positive potential can be used as a characteristic of emotional type recognition based on EEG signals.In order to further illustrate this,this paper also extracted three kinds of the mood of the timefrequency characteristics of EEG signals as a contrast to analyze,the emotion recognition accuracy of late positive potential characteristics is higher than the emotion recognition accuracy of time-frequency characteristics.It can prove that late positive potential can identify positive? neutral?negative mood of EEG signals,and it provides a way of identifying EEG signals emotion types.