Design And Research Of The Multi-target Brain Computer Interface System Of Visual Evoked Potentials Based On Code Modulation

Brain-computer interface(BCI)is a new communication system which establishes the connection between human brain and a computer or other electronic devices without the involvement of conventional peripheral nerve and muscle tissue.In recent years,the BCI research based on visual evoked potentials(VEP)has become more and more popular,among the various BCI modes based on VEP,the VEP BCI based on code modulation has the advantages of large number of optional targets and high information transfer rate.In this paper,we research the visual BCI based on coded modulation,and propose a new c-VEP BCI paradigm based on different codes grouping modulation,which can increase the number of stimulus targets on a large scale.In this paper,we design a multi-target brain-computer interface system based on c-VEP,which increases the number of stimuli to 64,and improves the design of visual stimulator on the basis of previous research.All targets are divided into four groups,which use different codes to modulate the stimulus targets and template matching method is adopted to identify the attended target.Eight subjects were invited to participate in the BCI experiment,and the experimental data were analyzed in detail to verify the feasibility and effectiveness of the system.Firstly,the correlation of the four reference template signals is analyzed.The results show that the stimulation response(c-VEP)signal generated by the four modulation codes used in this system has good auto-correlation and cross-correlation properties,making it easy to distinguish targets within a single group and between groups.Then the data of a single trial are classified using template matching.The results show that the classification accuracy rate of the BCI system is 88.36%,and the information transfer rate is as high as 184.6 bits/min.It is proved that increasing the number of stimulus targets can significantly improve the information transfer rate of the system.Finally,the data are deeply analyzed.The results show that the highest information transfer rate can be obtained by using the data of one stimulation cycle length to classify the target.The original EEG signal is band pass filtered in the frequency band of 2~30 Hz,and the spatial filter generated by the CCA optimization is used to spatially filter the multichannel c-VEP signals,so that the signal-to-noise ratio of the c-VEP signal can be greatly improved,and the highest classification recognition rate can be obtained.