Method for EEG Feature Extraction Based on Morphological Pattern Spectrum

Abstract

In order to classify the mental tasks in Brain-Computer Interfaces(BCI), a feature extraction method based on morphological pattern spectrum is here proposed. Flat morphological structure element is selected according to the characteristics of electroencephalography(EEG) and morph-ological features of different scales are obtained with pattern spectrum. Then, support vector machines(SVM) is used as the classifier. Testing results show that the average classification accuracy is up to 97.7% for two kinds of mental tasks and 93.0% for five kinds of mental tasks. This method has a simple calculation and effective feature extraction performance, so it could be a valid method for real time control of EEG.