A combined feature extraction method for left-right hand motor imagery in BCI

Author

Jie Hong ; Xiansheng Qin ; Jing Bai ; Peipei Zhang ; Yan Cheng

Author_Institution

Sch. of Mech. Eng., Northwestern Polytech. Univ., Xi´an, China

fYear

2015

fDate

2-5 Aug. 2015

Firstpage

2621

Lastpage

2625

Abstract

The aim of BCI is to translate brain activity into a command for a computer. For this purpose, the EEG signal processing plays an important role, especially in feature extraction. In this paper, a combined feature extraction method is proposed for left-right hand motor imagery in BCI. The power in the sensorimotor rhythm band and the statistical features of wavelet coefficients are used for extracting features and support vector machine is adopted for pattern recognition of left-right hand motor imagery. The performance is tested by the EEG signals of subject b and subject g from the datasets1 BCI Competition IV. The results have shown the availability of this method. It provides a novel way to EEG feature extraction in BCI.

Keywords

brain-computer interfaces; electroencephalography; feature extraction; medical signal processing; statistical analysis; support vector machines; wavelet transforms; BCI; EEG signal processing; brain-computer interface; electroencephalography; feature extraction method; left-right hand motor imagery; pattern recognition; sensorimotor rhythm band; statistical features; support vector machine; wavelet coefficients; Discrete wavelet transforms; Electroencephalography; Feature extraction; Pattern recognition; Rhythm; Support vector machines; Wavelet coefficients; BCI (Brain-computer interface); ERD/ERS; Motor imagery; SVM (Support vector machine ); Wavelet decomposition;

fLanguage

English

Publisher

ieee

Conference_Titel

Mechatronics and Automation (ICMA), 2015 IEEE International Conference on

Conference_Location

Beijing

Print_ISBN

978-1-4799-7097-1

Type

conf

DOI

10.1109/ICMA.2015.7237900

Filename

7237900