A BCI System Classification Technique Using Median Filtering and Wavelet Transform

Abstract

The brain–computer interface (BCI) system allows us to convert brain activity into meaningful control signals. This article presents an efficient BCI signal classification technique that uses median filtering and wavelet transform (WT) to improve classification performance and reduce computational complexity. In one preprocessing step, median filtering is carried out in order to attenuate noise, and WT is used to extract features that are classified by support vector machines (SVM). The database we use for this purpose is from BCI competition-II 2003 provided by the “University of Technology, Graz.” We show that using these two techniques in series, the classification accuracy can be increased up to 90 %. This method is therefore a very good approach toward designing online BCI and it is not computationally intensive.