Ensemble Classifier Model with Crow Search Feature Selection for Recognition of EEG Motor Imagery

Abstract

Abstract: The patient with limb dyskinesia can do his daily activities and rehabilitation training with the help of Brain Computer Interaction (BCI) based on Electroencephalography (EEG). But the EEG feature extraction and classification faces the issues of low efficiency and accuracy due to large individual differences and low signal-to-noise ratio. To solve this problem, this paper proposes a recognition method of motor imagery EEG signal based on ensemble classifier. To increase the quality of EEG signal characteristic data, this method initially uses the short-time Fourier transform (STFT) and continuous Morlet wavelet transform (CMWT) by pre-processing the collected experimental datasets, which is based on time series characteristics. In order to achieve high quality features, ensemble classifier effectively recognize the EEG signals. The quality of EEG signal feature acquisition is improved by ensuring the high accuracy and precision of EEG signal recognition. Haralick features are extracted to improve the motor imagery, where Crow Search Algorithm (CSA) is used to select the optimal features. The experiments are carried out on laboratory measured data and BCI competition dataset. The results showed accuracy of this method for EEG signal recognition is better than existing techniques.

Index Terms: Brain computer interaction (BCI), short-time Fourier transform (STFT), continuous Morlet wavelet transform (CMWT), Crow search algorithm (CSA), Horlick features and Ensemble classifier.