Matching pursuit for p300-based brain-computer interfaces

A correctly preprocessed electroencephalographic signal is necessary for building successful brain-computer interfaces (BCIs). Since the evoked responses obtained by stimulation are much weaker than the continuous EEG signal, the correct signal analysis enhances stimulation-driven signal components. This paper proposes methods for event-related potential processing for BCIs based on matching pursuit. The suggested method is compared with other simple methods which are frequently used for feature extraction. A multilayer perceptron was used for classification. The results can be used to improve the feature extraction for BCI systems.