Research On EEG Signal Recognition Method Base On SSVEP Paradigm And Jewelry Matrix Game

The development of brain computer interface(BCI),aiming to provide a communication channel from human to computer,directly converts brain activity into a series of control commands.In recent years,the application of BCI and computer game interactive is becoming more and more popular.Compared with other types of BCI game systems,the BCI game system based on the steady-state visual evoked potential(SSVEP)has higher recognition accuracy and shorter training time,which is of great significant in the aspect of promoting the practical application of BCI game system.However,there is some gap between the effect of SSVEP signal analysis and actual demand in a few channels and short time window situation.Therefore,this paper mainly discusses the research of EEG signal recognition method combining SSVEP paradigm and Jewelry Matrix Game,aiming to study a new method to improve the accuracy of SSVEP signal recognition method under the condition of fewer channels and shorter time window.Firstly,the recognition accuracy of SSVEP signal in a short time window to be improved on the premise of the less time cost and calculation cost.Canonical correlation analysis(CCA)has been proved to be a fast method without calibration data,so this paper hopes to improve the CCA method,and seek to improve the recognition accuracy of SSVEP signals without training data or calibration data.Therefore,a delay embedding into canonical correlation analysis(DCCA)method is proposed to embed the 1-order delayed version of the EEG signal when calculating the correlation coefficient.It has been proved that this method is better than CCA in the average recognition accuracy of SSVEP signals for all subjects under different time windows.Then,it focuses on improving the recognition accuracy of SSVEP signals in a short time window.The task Related component analysis(TRCA)method focuses more on the maximizing the covariance between trails,without considering the structure and characteristics of SSVEP signals.For this,this paper used the temporal characteristics of SSVEP signals to extend TRCA,proposed a task-related component analysis with temporal characteristics(TTRCA)method based on the time characteristics of SSVEP to seek a new set of spatial filters,and combined with template signals to further improve the recognition accuracy of SSVEP signals.The results showed that the average recognition accuracy of SSVEP signals of all subjects is better than that of TRCA in different time windows.Finally,the different time window lengths were selected to compare the two new methods proposed in this paper with the existing SSVEP signal recognition methods based on the SSVEP signal dataset of 14 subjects,and a systematic chart analysis was carried out to verify the feasibility and accuracy of the two new methods,and the application angles and suggestions of the two new methods were given.