Research On Emotion Recognition Based On EEG

How to accurately interpret the state of human emotion has always been a research hotspot in the field of artificial intelligence,medicine and computer.Emotion recognition is a research hotspot in the field of human-machine interaction.It can make computers more intelligent and human.Communicate and improve the user experience.Traditional emotion recognition technology is mainly based on speech and facial features.Because of love The external expression of the mood is easy to be disguised,so the traditional emotion recognition technology is not reliable.EEG based emotion recognition technology can make up for the deficiency of traditional methods,because EEG can reflect the electrophy siological activity of brain when processing emotion.Electroencephalogram(EEG)is a brain cell group activity recorded by the electrode,which is weak,only micro volt,and is very easy to be disturbed by data drift,eye electricity and electrocardiogram.Therefore,the denoising of EEG data is a very important research in the brain computer interface.Although the traditional singular value decomposition(Singular Value Decomposition,SVD)can remove the noise in the EEG signal to some extent,it is difficult to reconstruct the EEG signal because it is difficult to select the appropriate singular value.In this paper,based on the traditional singular value decomposition(SVD)EEG de-noising method,the sensitive component of EEG signal is selected by the sensitive factor and the location factor locates the corresponding singular value to reconstruct the time frequency signal of the EEG,so as to remove the noise in the signal and extract the effective EEG signal.Using the Hilbert yellow transform HHT(Hilbert Huang Transform,HHT),the instantaneous amplitude is calculated,and the average instantaneous energy characteristics are obtained.On this basis,the support vector machine(Support Vector Machine,SVM)is used for classification,and the average classification accuracy is obtained.The experimental results show that the improved SVD is more accurate than the traditional SVD classification algorithm,and proves that the improved SVD algorithm can effectively remove the noise in the EEG signal.We can recognize and process EEG only if we have the characteristic quantitythat can represent the emotion of the target.The same is true of emotional recognition based on EEG signals.In order to explore which kind of EEG characteristics have a higher classification accuracy rate for the classification of emotion recognition,this paper extracts 6 time domain features and 5 frequency domain characteristics of emotional EEG,and uses the t-test method,entropy based separability criterion,and probability distribution based discriminability criterion to select three methods.The SVM SVM is used to classify the EEG signals.The average classification accuracy of EEG characteristics in time domain and frequency domain was compared.Which EEG features can better identify target emotions? The experimental results show that the fused EEG characteristics are better than the single EEG classification,and there is no significant difference in the classification accuracy of the three feature selection methods.