Feature Extraction and Classification Algorithm of Brain-computer Interface Based on Human Brain Central Nervous System

Abstract

For the problems that brain-computer interface is susceptible to environmental noise interference and has low classification accuracy in traditional single-mode electroencephalography (EEG), this paper introduces the technique of function Near Infrared Spectroscopy (fNIRS) based on the study of EEG brain-computer interface in human brain central nervous system, designs and simplifies the experimental paradigm of the EEG-fNIRS multimodal brain-computer interface based on the fisting action. According to the result of single modal feature classification, the fusion feature based on EEG wavelet coefficient and fNIRS slope is proposed. The result shows that the average recognition rate of the fisting action task after feature fusion is 3% to 9% higher than that of the EEG feature and fNIRS alone. The fNIRS can significantly enhance the brain-machine interface performance based on the EEG and is of significance for the improvement of the application of the multimodal brain-machine interface.