The 7th International Conference on Computer Engineering and Network (CENet2017)

37 / 2017-05-13 19:42:05

Classification of EEG Signal by STFT-CNN Framework: Identification of Right-/left-hand Motor Imagination in BCI Systems

Electroencephalogram,Short-time Fourier transform ,Convolutional neural network,Motor imagery

This paper described the relationship between EEG signals and MI in BCI system. EEG signals were used to classify two kinds of direction of motion, left and right. We extracted features from original EEG data using STFT and put them into CNN, The result shows that the framework of STFT-CNN has higher average test accuracy. Furthermore, the generations of motor imagery were analyzed, the result shows that the better classification results will appear during the middle stage, and its classification accuracy could reach 92.8%.