Analysis And Recognition Of Eeg Signal In Various Emotional Processing

The research of emotion has been the research highlight in cognitive science,psychology,information science and other interdisciplinary studies.Particularly with the rapid development of brain science and information science,the technology of emotion recognition based on brain signals makes “reading emotions” become possible.Based on the advantage of high time resolution,noninvasive and portability,the technology of Electroencephalograph(EEG)becomes an efficient way to get emotion-related brain signal,and an objective and accurate way to analyze emotion.The emotion is induced by an internal or external stimulation by physiological or psychological activity.In different psychological experiment tasks,the processing mechanism of emotion is different,which can be correspondingly reflected different in some degrees of EEG signals in the human brain.In order to improve the recognition accuracy and practicality of EEG-based emotion recognition,this paper explores the neural mechanisms of emotion processing under different experimental tasks,seeks and analyzes of the relationship between emotion and the characteristic of EEG signals,builds corresponding emotional calculation model according to different emotions related psychological experimental paradigm.The main work is included as follows:1.Aiming at the problem that the ERP(Event-Related Potential)components,induced by Oddball experiment with emotion elicitation,are poor validity and aliasing,the key ERP components extraction method based on the spatial filter is proposed which can recognize the emotions with few trials.Due to the low signal-to-noise ratio of emotional ERP components,the ERP components are often submerged in the background EEG signals and aliased,a large number of trials are needed to enhance the ERP components which limit to the practical use.Therefore,on the basis of deep analysis for the ERP components characteristic,the key ERP components is extracted by constructing spatial filter.Combined with linear discriminate,at the same time,the maximization characteristic difference of diverse emotion ERP components are analyzed which realize emotion recognition by less number of experiment.Experiments demonstrate that the highest accuracy can up to 100% by averaging ERP components with only five trials,and the accuracy in single-trial is improved.Compared with the traditional method,the proposed method has higher accuracy and better practicability.2.Aiming at the limitation of the ERP components analysis in temporal and frequency which hard to fully characterize the emotion-related EEG features,a fusion method of ERP components and EEG features is proposed to emotion recognition.In the process of emotion elicitation,the ERP components reflect the spatial-temporal character of the neural signal while the EEG reflects the rhythms character.Due to the differences in the emotional processing with the two characters,only one of them is used in the previous studies which affect the validity of analysis.This paper proposes an emotion recognition method combined spatial-temporal characteristics of ERP components and EEG rhythm,including the temporal,spatial and frequency features.And the key EEG rhythm feature and key spatial features(channels)are analyzed.Experimental results indicate that the accuracy of the propose method has significant improve about 21% and 13% than the ERP feature or EEG rhythm feature used only.3.Aiming at the problem that lacking of object evaluation method on the cognitive effect of subliminal affective stimuli,a method to evaluation the subliminal affective priming effect based on the EEG signals is proposed.The subliminal affective priming experiment is a common paradigm in the current researches on cognitive effects of subliminal affective in psychological experiment.However,the traditional evaluation methods are based on the questionnaire or scale and other behavior which are non-objectivity.Therefore,based on the classical paradigm,the EEG signals induced by the subliminal affective are analyzed,Then the emotion recognition method is introduced to verify the priming effect.Besides,the differences between upper-liminal and subliminal emotion are compared.Experiment results demonstrate that the level of subliminal affective for positive and negative is significant higher than the random level of 50%(p<0.05).This method provides an useful approach for evaluation the subliminal affective priming effect in an objective way,and also the basis of the emotional processing mechanism in upper-liminal and subliminal elicitation.