BCI classification based on signal plots and SIFT descriptors

Brain Computer Interfaces are a challenging technology with amazing prospects but its push into mainstream assistive applications has not arrived yet. In this work a new method to analyze and classify EEG, Electroencefalography, signals, is proposed which is based on the extraction of visually relevant feature descriptors from images of the signal plots. This procedure has the advantage that the features which are used to classify are visually relevant and meaningful to a human observer, particularly to a physician, improving close collaboration and clinical adoption. Moreover, this may allow to tackle this demanding technology from a different perspective and improve the prospects of the BNCI, Brain/Neural Computer Interaction field.