Research On Game Control Based On Motor Imagination Eeg

Game is a new industry with rapid development and continuous popularity.Game,online game in particular,has been an industry with great concentration of technology and profits while 4G and 5G network spread.An unprecedentedly bright development prospect lies ahead in China because of large population.Playing game is a complex course which not only needs the coordinated control of the limbs,but also acquire the cognition,judgment and response of the brain to direct operation of players.Research on game control is mainly focused on motion track recognition,behavior tree,new human-computer interaction and other technologies up to present and the research on the application of EEG signal to game control has been reported rarely.Therefore,starting from the relationship between brain activity and game operation,this paper studies and establishes a game control system based on motor imagination EEG signal,aiming to improve the operability of the game and expand the means of human-computer interaction.In this paper,the preprocessing,feature extraction and classification algorithms of EEG signals of motor imagery are comprehensively combed and compared and optimized.Ten healthy volunteers were selected for motor imagination experiment which contains three kinds of motor imagination scenarios:left hand movement imagination,right hand movement imagi-nation,foot movement imagination and blank control.Then we collect,preprocess,feature-extract and classify the EEG signals of each subject.Through optimization and screening,a game control system based on motor imagination EEG signal is established.And we design and implement a game and use our control system in it.It turns out our control system works well.The main results of this study include:(1)We optimize and improve the traditional CSP and use it to extract features of motor imagination EEG which was preprocessed already.It turns out that using optimized CSP to extract features of the three kind of MIEEG makes the accuracy of classification higher and more steady.It is conducive to further research in the later period.(2)Three classification methods,linear discriminant analysis,support vector machine and Naive Bayes classifier,were used to classify the EEG data of 10 subjects in this study.And the accuracy of the three classification methods was compared and evaluated.The results showed that the accuracy of SVM and Naive Bayes classifier was more than 90%,significantly higher than that of LDA.(3)A dodge game was designed based on the control of motor imagination EEG,and it was tested by five volunteers.The results show that the game runs smoothly and all of the five volunteers can successfully control the game with motor imagination EEG.However,with the increase of the difficulty of the game,the failure rate of EEG signal control increases,and further optimization of the control system is needed,especially in the algorithm of feature extraction.