Study On High-Performance Online Motor-Imagery Brain-Computer Interface Systems With OpenVIBE

Brain Computer Interface(BCI) can be a way transmitting information based on EEG that independent of muscle. BCI has many great features like low cast, high operability and high safety. These features can bring new lifestyle to disabled person and help them to shorten the rehabilitation period. Motor imagery Brain Computer Interface based on the physiological phenomenon called ERD/ERS is becoming a great hot research direction on BCI now. The study on High-Performance Online Motor-Imagery Brain-Computer Interface System has important significance for us.At present, many advanced theories about Motor-Imagery Brain-Computer Interface continuously have been raised to improve its performance to a certain extent. However, the current BCI systems are mostly designed by the self of BCI laboratory, although the systems can meet the requirements of experiment, but there are also many shortcomings in them. For example, the poor repeatability that is not easy to study and exchange in different teams; the poor readability that can’t directly reflect the role of the different parts of the program, and the experiments need to have some understanding of the program, and so on. The design of the BCI system is based on OpenVIBE, a kind of experimental platform. With its unique modular operation we can greatly reduce the difficulty of the operation of the laboratory personnel and improve the efficiency of the experiment.In this paper, it based on regular temporal filtering classification algorithm(RSTFC), designed and implemented a High-Performance Online Motor-Imagery Brain-Computer Interface System. The algorithm of RSTFC that is based on a famous algorithm called Common Spatial Pattern can optimize spatial and temporal filters at the same time, and it also trains the sparse classifier on this basis to pick out the most separable feature. The algorithm has the feature of small amount of computation and fast execution speed. Compared to the existing Motor-Imagery Brain-Computer Interface System, this paper presents a new system that significantly improves the online classification accuracy of Motor-Imagery and it also has good scalability and great prosects for practical.