期刊名称：International Journal of Computer Science Issues
印刷版ISSN：1694-0784
电子版ISSN：1694-0814
出版年度：2019
卷号：16
期号：1
页码：1-7
DOI：10.5281/zenodo.2588239
出版社：IJCSI Press
摘要：In this research, we present a filtering method for cleaning the EEG signal based on empirical mode decomposition (EMD) to enhance classification accuracy associated with upper limb real movement. In our method, we decompose each channel of EEG signals into a set of IMFs using EMD. We select or reject IMFs based on a calculation of dominant frequency of every IMF. In this procedure, we reject some IMFs whose dominant frequency greater than 30Hz. After reconstruction of the signal, we apply common spatial patterns to reduce the dimension as well as find out features of the signal and then we use support vector machines to classify left hand and right hand movement. Our paper demonstrates that the proposed method successfully extract features from row EEG signal that carries more information by filter out irrelative information. Our method tested on a publicly available dataset and obtained a significantly better performance.
关键词：Brain computer interface; Empirical mode decomposition; Dominant frequency analysis; Common spatial pattern and Support vector machines.
