Classifying Motor-imagination Signals in Brain-computer Interface Based on Feature Extraction of Parametric AR Model

In the study of the brain-computer interface algorithm, this paper analyzes the feature extraction and pattern classification algorithms of left and right hand motor imagination EEG signals, uses the AR parameter model to extract the features of the signals and compares them with the features extracted by wavelet transform. Extract the EEG signals of C3 and C4 channels for analysis, use the Burg algorithm to extract the 6th-order AR model coefficients of the EEG signals; use Daubechies wavelet functions to decompose the EEG signals into three layers. The third layer corresponding to the motor imaging EEG signals Wavelet transform coefficients are used as features. The support features are used to classify the extracted features. Finally, the results show that the features extracted by the AR parameter model method are better than those extracted by the wavelet transform method, and the classification accuracy is higher.