Abstract

Info

Title:: Recognition of motor imagery EEG signals based on multi-feature fusion

Author(s):: JIANG Yue; ZOU Renling; School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China

Keywords:: ; Keywords: electroencephalogram recognition; feature fusion; principle component analysis; support vector machine; motor imagery

Abstract:: Abstract: Objective The external devices are controlled with brain computer interface after the detection of electroencephalogram (EEG) signals. A feature extraction method based on multi-feature fusion is proposed to solve the problem that the traditional method of single feature extraction cannot realize the multi-angle characterization of EEG. Methods The initial eigenvectors of time-frequency-space domain were extracted by autoregressive model, empirical mode decomposition and common spatial pattern, separately. Subsequently, principal component analysis was used to reduce the dimension. Finally, support vector machine is used to classify the motor imagery EEG signals. Results After the data processing of BCI2003, the recognition rate reached 91.9%, higher than that obtained by the extraction based on single feature and the combination of any two features, and that obtained with BP neural network and probabilistic neural network. Conclusion Feature extraction method based on multi-feature fusion can characterize EEG better, and the combination with support vector machine can achieve better classification results, which proves the effectiveness of the combined use of multi-feature fusion and support vector machine. The proposed method can be further applied in brain computer interface.