EEG signal feature reduction and channel selection method in hand gesture recognition BCI system

With the purpose of determining the best feature set of BCI system based on EEG, this paper proposes a feature dimension reduction method on the basis of Pearson’s correlation coefficient. Feature extraction time and classification accuracy are used as evaluation criteria by using SVM classifier. Feature dimension is reduced from two aspects of feature type and number of channels. Time domain, frequency domain, time-frequency domain and spatial domain features are extracted for comparison in the designed hand gesture recognition experiment. Compared with the running time and classification accuracy of PCA and LDA algorithms, it is confirmed that the feature dimension reduction algorithm based on Pearson correlation coefficient can effectively reduce the feature extraction time and improve the classification accuracy, and obtain the most suitable system feature subset.