Design And Research Of Brain Computer Interface System Based On VEP

Brain-computer interface(BCI) is the new communication and control method that realizes direct communication between human brain and external device by analyzing electroencephalogram signals, and is independent of peripheral nerves and muscle. In recent years, the practical application of BCIs draws increasing attention from researchers. With a goal of practicability of BCI, two test platforms based on frequency modulated visual evoked potential(f-VEP) and code modulated visual evoked potential(c-VEP) were set up respectively, and on which two character input systems were realized.The BCI system based on f-VEP includes a visual stimulator, a signal acquisition module, and a real-time signal processing module. The character input method of decision-tree and the signal processing method of canonical correlation analysis(CCA) are employed to recognize stimulus targets by searching the maximum correlation value between the steady-state VEP(SSVEP) signal and reference signal composed of sinusoidal components of stimulus frequency and its harmonics. The system performance was tested by off-line and on-line experiments. The classification accuracy and information transfer rate(ITR) of 95.8% and 27.2bits/min were achieved respectively. Compared with the traditional SSVEP based character input systems, the decision-tree based character input system greatly improves the practical performance of BCI based character input systems.A character input system based on c-VEP was designed,in which the visual stimulator is modulated by Golay complementary sequences. The CCA method and template matching method are used to identify stimulus targets. Compared with the system which is modulated by M sequence, the character input system based on Golay complementary modulation yielded higher classification accuracy and information transfer rate through the off-line analysis and online verification. Thus the c-VEP BCI system further enhances the applicability of a character input system.