| Brain-computer interface(BCI) is a novel way of interaction between a human and a machine, which directly links the brain and the outward environment. In this thesis, we proposed a novel multiple inputs BCI protocol utilizing a great number of simultaneous chromatic visual stimuli modulated by orthogonal pseudo random sequences and a realization of such protocol was tested on normal healthy human subjects. According to properties of human visual system, chromatic rather than traditional achromatic visual stimulation has been demonstrated much safer and more comfortable for BCI users. For the choice of psychophysical parameters of the chromatic visual stimuli optimal for BCI, it was found through systematic experiments that strong visual evoked potentials(VEP) could be elicited for subjects sitting 30 cm away from the screen(a visual angle of 12 degree) under a lab environment of dim light close to the screen lighting background.In the BCI protocol proposed in this thesis, a total of 36 sequences of chromatic con-central rings were simultaneously presented on and off a computer screen to simulate a 36 alphanumeric keyboard input system, with the on/off pattern each specified by one of 36 orthogonal pseudo random Gold sequences. Command to the BCI system was determined by which one of the 36 chromatic stimuli the subject focused on. Since the Gold sequences are orthogonal to each other, the measured EEG signals of the subjects were processed by the according decoding algorithm to obtain the VEP signal elicited by each and all of the chromatic ring stimuli. Then the decoded VEP signals were averaged over multiple trials and filtered by a matched filter using a template of clean VEP signal of the subject to improve signal-to-noise ratio(SNR), resulting improved accuracy of BCI commands detection. Experimental results showed that the proposed BCI system could attain an average information transfer rate(ITR) of 21.06 bit/min at an average command detection accuracy of 91.3%。... |
PDF Full Text Request