The Influence Of See-through Display And Visual Information Overlapping On Brain-computer Interface And Methods For Improving

Brain-computer interface(BCI)and augmented reality(AR)provide new ways of information interaction,and the technical advantages make these two technologies arouse wide concern.The combination of these two technologies will provide a new reliable interactive way to meet the high-performance application requirements in some special fields as aviation and spaceflight,which contains the interaction among humans,machines and the environment.BCI can provide more flexible and non-physical interaction for AR,and AR can provide a new way of display for the application of complex environment for BCI.But in the AR display environment,seethrough display(STD)and visual information overlapping(VIO)bring a decline in BCI output performance.To research the influence of STD and VIO on the performance of BCI,and find the corresponding improvement methods,this paper carried out the corresponding experimental research through offline and online system construction.Firstly,P300 and SSVEP were chosen to test the performance change of BCI system under the STD and VIO condition.Linear discriminant analysis(LDA)and four kinds of classification algorithms based on canonical correlation analysis(CCA)were used in the online and offline experiment,which contained 10 subjects.The result showed that STD brought the decline of classification accuracy of 7.00% in P300 and 3.38% in SSVEP.And when the background interference was added to the experiment by VIO,the decline of classification accuracy was up to 7.70% in P300 and 25.63% in SSVEP,and the difference in SSVEP showed significance of P<0.001,revealed the serious impact of STD and VIO on the performance of SSVEP-BCI and P300-BCI.Then,this paper tried to find the way of improvement on SSVEP,and explored the system performance promotion scheme respectively on the visual evoked mode and instruction output strategy.By the comparison experiment of 4 visual evoked mode,based on 5 subjects,the mode of 4cm*4cm solid box flashes brought the most noticeable rise of classification accuracy of 17.60%.After that,an output strategy was established based on posterior probability threshold.Combining the output strategy with the visual evoked enhancement scheme,an online system with information closedloop was built.The online system realized stimulus modulation based on decision information.And with the system,the online and offline experiments were designed.8 subjects participated the experiments,and got the performance improvement of 15.00% offline and 22.19% online.This paper studied on the influence of STD and VIO on the performance of BCI,the performance improving methods and the online platform will provide an important theoretical and technical basis for the reliability and the optimization of environmental adaptability of AR-BCI system.