Research On EEG And Eye Movement Emotion Recognition Method

There are many ways to express human emotions.In recent years,with the improvement of the level of the electronic technology manufacturing industry and the continuous advancement in the field of artificial intelligence,it has become a popular direction of current research to collect a variety of human physiological signals for multi-modal emotion recognition.However,because some physiological signals are not easy to collect,and there are differences between different people's emotion expressions,the effect of emotion recognition is low.In response to these two problems,this paper has carried out research on the method of using eye movement signals and EEG signals for emotion recognition.The main work of this paper is as follows:An emotion recognition scheme based on eye movement is proposed,which adopts a novel emotion recognition method based on pupil position.The pupil position of this scheme is obtained from the facial video of the multi-modal physiological signal emotion database(DEAP),which mainly uses the gradient-based pupil position localization algorithm.For pupil position data,this paper proposes a comprehensive waveform complexity.By segmenting the pupil position coordinate waveform,the sum of all waveform segment correlation coefficients is used as the eigenvalue,and it is compared with Lempel-Ziv complexity and sample entropy The experimental results show that the proposed comprehensive waveform complexity emotion recognition rate is the best,with a recognition rate of 58.56% in the Valence dimension and 69.72% in the Arousal dimension.An emotion recognition method based on fuzzy integration of eye movement and EEG multimodal fusion is proposed.In this scheme,for the EEG signal,the electrode channel selection based on the emotion generation mechanism is adopted,and the differential entropy is used as the feature value for the EEG signal of the selected electrode channel,and the emotion recognition result of the EEG signal is obtained through LIBSVM.Then the fuzzy integral is used to fuse the results of the eye movement emotion recognition and EEG emotion recognition in the previous scheme.In the experiment,after fuzzy integration,the emotion recognition effect of the fusion of eye movement and EEG modalities reached 75.26% in the Valence dimension and 78.64% in the Arousal dimension.An emotion recognition scheme based on Elliott waveform theory and neural network is proposed.The scheme is based on Elliott waveform theory,segmenting the pupil position coordinate waveform,and then using BP neural network to perform pattern recognition on the pupil position coordinate waveform,and the waveform classification results are stored as a knowledge base;then,the pupil position waveform recognition results are sent into In the prediction classification neural network,the emotion corresponding to the waveform is recognized.The experimental results show that the effect of the second classification in the Valence dimension reaches 72.5%,and the effect of the second classification in the Arousal dimension reaches 74.24%.