Chinese Journal of Tissue Engineering Research ›› 2011, Vol. 15 ›› Issue (48): 9007-9010.doi: 10.3969/j.issn.1673-8225.2011.48.019

Previous Articles     Next Articles

Features extraction of brain-computer interface based on adaptive autoregressive models

Wang Jiang1,2, Xu Gui-zhi1, Wang Lei1, Zhang Hui-yuan2   

  1. 1Electrical Engineering School of Hebei Technology University, Tianjin  300130, China
    2Tangshan Vocation and Technical College, Tangshan  063000, Hebei Province, China
  • Received:2011-05-30 Revised:2011-07-18 Online:2011-11-26 Published:2011-11-26
  • About author:Wang Jiang★, Master, Lecturer, Electrical Engineering School of Hebei Technology University, Tianjin 300130, China; Tangshan Vocation and Technical College, Tangshan 063000, Hebei Province, China tswxwj@yeah.net
  • Supported by:

    the Natural Science Foundation of Hebei Province, No. E2009000062*

Abstract:

BACKGROUND: Due to the non-stationary nature of electroencephalograms (EEG) signals, brain-computer interface (BCI) cannot walk out of the laboratory. Due to dynamical changes of the EEG signals under the disturbance of physical or psychological conditions, it is difficult to obtain stable classified features, which is one of the main reasons for the practical utility of BCI.
OBJECTIVE: To observe the classified features of EEG signals base on left hand, right hand and foot movement in order to improve the reaction rate and accuracy of classification in BCI
METHODS: Three subjects were selected for the BCI experiment. Three mental tasks with the imagination of left hand, right hand and foot movement were followed according to the screen instructions. The disturbance of electro-oculogram in EEG signal was removed by using band-pass and Laplace filter. Model parameter of many variables adaptive autoregressive models (MVAAR) was extracted as the classified features. The comparison between MVAAR and traditional adaptive autoregressive models was made.
RESULTS AND CONCLUSION: The result showed that classified features of left hand, right hand and foot movement could be extracted stably in MVAAR, which is beneficial to improve the adaptive ability of online BCI and the application of BCI system.

CLC Number: