Mental tasks classification for BCI using image correlation

This paper describes a classifier based on image correlation of EEG maps to distinguish between three mental tasks in a Brain-Computer Interface (BCI). The data set V of BCI Competition 2003 has been used to test the classifier. To that end, the EEG maps obtained from this data set have been studied to find the ideal parameters of processing time and frequency. The classifier designed is based on a normalized cross-correlation of images which makes possible to obtain a proper similarity index to perform the classification. The success percentage of the classifier has been shown for different combinations of data. The results obtained are very successful, showing that this kind of techniques may be able to classify between three mental tasks with good results in a future online testing.