Emotion Recognition Research Based On EEG Signal

Nowadays,Researches about brain signal has became a hot topic in brain machine interface(BCI),artificial intelligence(AI),computer science field and health medical et al.in which the emotion recognition research based on EEG signal is increasingly hot,either take emotion recognition to application of artificial intelligence,or assist the research and treatment of mental diseases like depression and autism.Emotion recognition research based on EEG signal have significant practical significance and the application value.This paper introduced fractal characteristic to emotion recognition research based on EEG signal—take fractal dimension as features value of EEG signal to distinguish different emotional states,research is divided into three major parts,the first part,is to calculate the accuracy of fractal dimension to be eigenvalue about emotion recognition.The second,use fractal dimension eigenvalue to determine brain relevant area and frequency band about emotion states based on the research above.In the process about calculation of the recognition rate,principal component analysis(FCA)algorithm is selected to achieve dimension reduction analysis of 64 channels EEG signal,then take 6 channels Higuchi fractal dimension's average as eigenvalue.The results show that support vector machine(SVM)get the highest classification accurate rate 83.33% with the contrast of two kinds of classification algorithm.We understand that emotion recognition using a 64-channel EEG is not realistic in practice case,so it is imperative to reduce the number of signal channels.At the same time,in consideration of appearance about the rhythms of EEG signal,this paper choice HF as eigenvalue to search for brain area and frequency have great relevance to emotion states.Results showed that classification accuracy higher in frequency bands like alpha,beta and gamma band obviously exceed lower frequency range,and the value of HF from frontal and temporal lobe is greater than others.At this point,it can prove that the fractal characteristics of EEG signal can be used for emotion recognition.