Study On Motor Imagery BCI Based On STM32and Labview

Brain-computer interface (BCI) is a kind of communication system that does notdepend on the normal output channels of brain. With the direct channel between brainand computer, BCI system can directly convert the information from brain into thecommands to control some peripherals. Therefore, BCI provides a valuable creativecommunication alternative to mortal gestures or language, which helps incommunicating or controlling. With the tremendous demand for BCI system,Rehabilitation medicine, industrial, military and other fields promote brain computerinterface has become a hot subject. However, before we reached a better future, the BCIsystem widely spread, most of the world’s researh on brain-computer interface is still inthe laboratory stage.Facing the opportunity and challenge of BCI technology, the research of theapplication on real-time BCI based on LabVIEW is carried out in this paper. The BCIsystem constructed to process motor imagery EEG. Compared to visual evokedpotentials, motor imagery EEG is easy to be detected. Due to the low frequency ofstimulation, the system is difficult to cause visual and mental fatigue. Compared toembedded systems, the computer has an overwhelming advantage in terms of hardwarecomputing speed. That’s why we choose the LabVIEW software which able to callcomputer resources fully to develop our system.According to the motor imagery brain-computer interface application requirements,we use LABVIEW software to design a new visual stimulation. The stimulate imagealternating in a particular frequency alternating. The vivid stimulating pictures caneffectively prompt experimenter to execute their left and right hand motor imagery. Inorder to guide training, stimulating interface can simultaneously feedback doctorsreal-time EEG waveforms.The most important study on BCI technology is to find the suitable algorithms toextract weak motor imagery EEG from strong background noise, and distinguish theexperimenter’s choose. Having compared wavelet decomposition, wavelet packetdecomposition and reconstruction, SVM and BP neuron network, finally we chose db5wavelet packet filtering to reconstructing EEG which in a specific time window. Thenselect BP neuron network to recognize feature vectors which come from sliding energywindow data, and using quantum particle swarm optimization to make BP network works best.Experiments illustrates that the system can effectively prompt the experimenterproducing motor imagery thinking movement. The choice of algorithms for offline datawith high recognition rate, validate the use of imagination with exercise trainingLABVIEW-based brain-computer interface is feasible.