Research On The Brain Cognition Rule And EEG Analysis Method For Image Emotion

Emotion is a comprehensive state that contains people's feelings,thoughts and behaviors.In human evolution and daily life,we can say that the role of emotion is everywhere.As early as nineteenth Century,there was a variety of methods used to study emotions.In recent years,the study of emotional cognition and its brain mechanism,especially the relationship between emotion and cognition and its underlying neural basis,has gradually become a hot research field in cognitive neuroscience.Electroencephalogram(EEG)is widely used in all kinds of emotion research because of its high resolution and synchronization of brain activity.The purpose of this thesis is to study the cognitive law of brain activity for the analysis and recognition of emotional images.It is devoted to the research of cognitive law of visual information and the technology of EEG signal processing under the guidance of the research of visual information cognition law.First,the image emotion recognition rules based on the International emotional standard picture database(IAPS)are designed.The Russel emotion classification model is used to divide emotions into two dimensions: pleasure and arousal.120 emotional images in IAPS were selected as stimuli to produce different emotions,to record EEG,and to establish an emotional picture combined with effective event related potential and behavioral experiments to induce cognitive neuroscience experimental paradigm of EEG.Then we study the brain cognition law of image emotion,and use ERP technology to divide into two aspects to explore the emotion cognition rules induced by human being processing visual information: the analysis of brain activity based on emotional titer and the analysis of brain activity based on emotional arousal,which can provide us with the effective discrimination of the emotional state of the image.The basis of neuroscience.We found that in the early stage of visual information processing(former 200ms),different processing of visual information of different titers had been processed;and this difference processing occurred in the frontal lobe and the parietal lobe,showing the symmetry of the left and right brain;in the middle and late stage of visual information processing(300-600ms),the difference was also reflected.The difference processing of visual information difference between different titers exists in every stage of mental processing,and the different processing modes of visual information difference between the left and right brain for different titers are different,but they all participate in the process of emotional processing.At the same time,we also found that the brain regions of different titers were overlapped,and the visual emotions of different titers were regulated by the common neural channels;the right hemisphere of the brain had a relative advantage in the process of emotional stimulation.At last,the neural network which combines the convolution neural network and the long and short time memory network is used to extract and classify the characteristics of EEG data.By using the cognitive rules obtained,the network structure and parameters are improved to improve the classification recognition rate.The average recognition rate of the 6 subjects' EEG data is increased by 9.23%,single name.The highest can reach 55.56% in the subjects.The research work of this paper has established the cognitive neuroscientific experimental paradigm of EEG induced by effective emotional images.The cognitive law of human processing emotion images is obtained by using ERP technology,and the recognition rate is improved by using it to extract the features of EEG data and improve the classification algorithm.This provides a new idea and an effective improvement scheme to solve the problem of EGG signal classification and recognition.It also has a certain promotion value.At the same time,it also provides a new way of thinking and direction for the application and development of the convolution neural network,such as the application and development of the deep learning method in the EEG signal processing.