Emotion Classification System Using VR Scenes Based On Flexible Textile Eeg Electrode

The development of 5G technology will accelerate the development of virtual reality applications.Virtual reality technology has the advantages of immersion,interactivity,and multi-perception,which can bring more real and strong emotional feelings.The wearer's emotional response is extremely important for the interactive application of virtual reality in cultural education,entertainment and other fields.The EEG-based emotion recognition method provides the possibility of wearer's emotion monitoring.Designing and researching a wearable virtual reality emotion classification system based on sparse channel EEG monitoring is of great significance for promoting the application of virtual reality technology in daily life.Therefore,this article will carry out in-depth research on the design of forehead EEG dry electrode and the establishment of virtual reality emotion classification system.There are several problems in the current research:(1)The existing EEG dry electrode has high contact impedance,low acquisition accuracy,and insufficient wearing comfort,which is not conducive to the integrated use in virtual reality devices;(2)In current research there is a lack of EEG data sets induced by standard virtual reality emotional materials,and there are few studies on emotion classification based on such data sets;(3)In the emotion classification experiment of EEG signals,there are large differences in the distribution of EEG signals of different individuals.The method has a low accuracy rate for the emotion classification of new individuals who do not appear in the training set,which limits the practical application of the virtual reality emotion classification system in daily life.Based on the above problems,this paper mainly carried out the following three tasks:(1)This paper designed a multi-layer flexible fabric EEG dry electrode,and carry out systematic experiments to verify the performance of the electrode.The contact impedance of the electrode in the forehead area is 8.97 ± 1.97k?(@ 10Hz),and the average correlation with the wet electrode is 95.9% ± 1.7%,and the electrode has good long-term stability and sweat absorption performance.The proposed electrode can achieve the same effect of wet electrode in forehead EEG collection,and can be effectively integrated into virtual reality equipment;(2)A virtual reality emotion classification system based on fabric electrode forehead EEG collection has been established,EEG data collection experiments induced by standard virtual reality emotion-evoked materials was conducted,and data of 18 subjects was accumulated.The time domain,frequency domain and spatial asymmetry features has been extracted based on the obtained data,and input them into the integrated decision tree model for emotion classification.The GBDT method reached an average accuracy of 82.34% in five experiments.By comparing with other public EEG datasets,the validity of the dataset established in this paper is verified;(3)Aiming at the problem of emotion recognition for new individuals,a deep domain adaptive method based on convolutional neural network is proposed.The maximum mean difference of the joint distribution(JMMD)is calculated in the fully connected layer of the network to reduce the distribution difference in different fields during the learning process.The average classification accuracy rate of this method on 18 subjects is 69.1%,which is nearly 10% higher than that of traditional machine learning methods.The accuracy rate on DEAP dataset is 70.3%,which is higher than the related research results.Based on the work carried out on the above three issues,this paper designed a new multi-layer fabric EEG electrode,formed a new virtual reality emotion classification system,and established a sparse channel EEG data set induced by standard virtual reality emotion-evoked materials,and provides an effective research method to solve the individual difference,which is beneficial to the application of the system in daily life.