EEG Pattern Recognition based on Frequency Analysis during Imaging and Gazing Tasks

Abstract

A brain-computer interface (BCI) is a translation system that detects human's intentions from EEG (electroencephalography) and converts it to computer commands. This system has been expected to help the disabled (e.g. amyotrophic lateral sclerosis or spinal cord injury patients) to communicate with the other people. In this paper, we tried to estimate shapes from EEG during tasks(gazing or imaging) shape in order to apply to the BCI system. The optimal combinations of parameters were investigated to discriminate shapes by using frequency analysis. Then features were used to recognize among three different shapes. As a result, accuracy of classification was more than 70 percent in some tasks. This fact suggests that a new type of BCI system will be available.