VEP Feature Extraction and Classification for Brain-Computer Interface

It is necessary to extract and recognize visual evoked potential (VEP) in the transient VEP based brain-computer interface research. Averaging method and wavelet filtering were used to extract the weak VEP signal noisy raw signal. Feature vectors gained in wavelet transform domain were input to the perceptron to achieve the recognition of VEP and produce the brain-computer interface control signal. Experiments showed that the feature extraction in the wavelet domain can effectively extract the VEP signal feature, reduce noise and decrease the dimensionality at the same time, and the VEP classification algorithms can recognize the VEP signal and produce the brain-computer interface control signal correctly, which is useful to increase the information transfer rate of brain-computer interface