Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
10344652 | Computer Methods and Programs in Biomedicine | 2005 | 7 Pages |
Abstract
Features can be classified into interferential features and discriminable features according to their contribution to pattern recognition. In this paper, a novel and simple method based on wavelet packet transform is proposed to extract the features from surface EMG signal. In this method, the features are relative wavelet packet energy (RWPE), which is evaluated from several selected frequency bands of surface EMG signal. Compared with a conventional method, which is of the best performance in previous applications, the method can compress the interferential features and enhance the discriminable features more effectively. In consequence, the RWPE features calculated by the method represent different patterns of surface EMG signal more accurately and the accuracy of surface EMG signal pattern classification is improved greatly.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science (General)
Authors
Xiao Hu, Zhizhong Wang, Xiaomei Ren,