Research of feature extraction of BCI based on common spatial pattern and wavelet packet decomposition

Brain-Computer Interface (BCI) is to establish a new communication system that translates human intentions reflected by EEG into a control signal for an output device such as a computer. This paper classified the EEG of two kinds of motor imagery. The feature extraction method combines wavelet packet decomposition and common spatial pattern. The k-nearest neighbors (KNN) is applied as classification method. The raw multi-channel EEG data is pre-processed by wavelet packet decomposition, with CSP method to extract the feature, and the best classification accuracy can reach 95.3%. If the EEG data is not decomposed by wavelet packet, the classification accuracy is only 83.3%. The result shows that if wavelet packet function and level is selected properly, the classification accuracy can improve effectively.