Research of classification methods of EEG signal based on wavelet packet transform and CSP

These years have been witnessing an increasing emphasis on researches of brain computer interface (BCI) which becomes a novel communication method from the brain to the output device, independent on normal peripheral nerve and muscle. And electroencephalogram (EEG) signal processing is one of the key research topics. In this paper, wavelet packet transform and common spatial patterns (CSP) are utilized for feature extraction. Finally, support vector machine (SVM) and Mahalanobis-distance are chosen to classify two kinds of motor imagery signal of left and right hands. Through experiments, we can recognize various factors affecting classification accuracy and the maximum accuracy rate could be up to 90.00%.