SRC analysis of EEG signal for activity detection

Abstract

In recent years, SRC has received many attentions for classification and identification tasks. This paper attempts to introduce a sparse representation based classification of EEG signal features and identification of associated activities or tasks. It uses wavelet and ICA processing of EEG signal for feature selection and dictionary training. Multiple dictionaries are trained and used for EEG signal encoding. Similar features are extracted from the test EEG signal and attempted to be approximated by using the atoms of trained dictionaries. Desicion is made in favour of the task corresponding to the dictionary from which maximum number of atoms are used in representing the features of test signal. On the basis of the results, we find that the proposed multi-dictionary scheme shows acceptable classification accuracy.