Research on EEG Feature Extraction and Recognition Method of Lower Limb Motor Imagery

Abstract

Aiming at the problems of difficult signal acquisition, low signal-to-noise ratio and poor classification accuracy of BCI technology, based on the theory of EEG, this paper designs a leg raising EEG experiment of lower limb motor imagery and collects EEG signal data from 20 subjects to improve the accuracy of classification and recognition The process of feature extraction and classification recognition is explored, and a multi domain fusion method is proposed for EEG signal feature extraction from time domain, frequency domain, time-frequency domain and spatial domain. At the same time, bagging and gradient boosting ensemble learning algorithms are applied to EEG signal classification and recognition, and multi domain fusion features are tested by constructing different classifiers, The final classification accuracy reaches 87.8% and 93%, which is better than the traditional SVM classification method.