The Brain-Computer Interface Research On Event Related Potential

Brain-Computer Interface (BCI) is a direct information communication and control channel established between human and computer or other electrical devices and it is a wholly new communication system that does not depend on the brain's normal output pathways of peripheral nerves and muscles. EEG-Based BCI may provide those 'locked-in' but with intact ideation, with an effective communication and control channels with outside world. That's why BCI is winning more and more attentions.The whole BCI System constitutes of Signal acquisition and pre-processing, feature extraction and classifying algorithm, output of controlling signal.At First, the neurological basis of EEG is generally summrized, especially for the characteristics of event-related potentials.Then it shows the parts of the BCI System and their aims, usual methods. Secondly, as the 2003 global BCI competition gave the competitiers two problem about Event-related Potential which request them to study processing and analyzing the signal data, I have done some research on the aspects of pre-processing, feature extraction and classifying algorithm.The major characteristics of P300 Signal is that the larger positive wave will be seen after 310 ms when the target stimulus accurs. In unnoisy evirionment, the positive wave reaches the maxmimum after the target stimulus accurs.In the P300 signal processing, the samples of every letter after target stimulus or non target stimulus accurs are gathered and denoised by adding these signal data. To Speed up the analyzing, different methods are used to ignore those channels which have little P300 element and impose little contribution on the classifying results.Finnally, the potential of useful channels is sampled, which constitutes of feature vector, to be classified with support vector machine and the correct classifying rate is up to Percent 87.24.The elements of mu and beta rhythm among the evoked potential are caused by imaging left hand moving or tongue moving, but they have different Power spectrum, frequency and more useful channels. So we choose these more useful channels according to event-ralated desynchronization phenomenon and then count the power value of different frequency of these channels.We portfolio information from the time zone and frequency zone into feasure vector to classify.It can obtain higher correct rate than the former methods by classifying with frequency feasure.At last, we compare the process of dealing two problems, point out the direction of further research and the measure of further enhancing the correct rate.