Classification of Motion Imagery Signal Based on Wavelet Packet Decomposition and Common Space Pattern Algorithm

Abstract:
The brain-computer interface system provides a solution for the paralyzed patients with motor disability, while the traditional general spatial pattern (CSP) extracts the features of electroencephalogram (EEG) signals only to extract the spatial domain information between the channels, but elide the information features of the signal in time domain and frequency domain to improve the classification accuracy of motion imagination EEG in brain-computer interface system. In this paper, we propose a feature extraction algorithm for motion imagination signal based on wavelet packet decomposition and CSP algorithm. The motion imagination μ rhythm and ß rhythm are extracted by wavelet packet decomposition, and the interference signal is deleted to the maximum extent by using the limitation of time window. CSP algorithm is used to filter spatial domain to extract features. With wavelet packet decomposition and CSP algorithm, the frequency domain and spatial domain information of the signal are fully utilized. In this paper, we design a complete process of filtering, processing and analyzing motion imagination signals from signals. The recognition rate of two kinds of motion imagination signals can reach 88.86%.