Mental Tasks Signal Classification

Abstract

Electroencephalogram (EEG) signal is an important source of information for knowing brain processes. To interpret the brain activity, Matching Pursuit Based EEG signal classification is proposed. This system includes three main components which are Preprocessing, Feature extraction and Classification. In the preprocessing step, Wavelet Packet Independent Component Analysis (WPICA) method is used to remove some unwanted noise of EEG recording. Matching Pursuit (MP) with Wavelet Packet Dictionary is used to extract the features of EEG signal. The k Nearest Neighbor (kNN) classified the extracted MP features. In this work, the Keirn and Aunon EEG dataset is used in the experiments. The feature extracted from MP based wavelet packet dictionary achieved over 90% accuracy in two seconds length of brainwave signal in five mental tasks classification.