东北大学学报（自然科学版）

东北大学学报(自然科学版) ›› 2010, Vol. 31 ›› Issue (10): 1377-1380.DOI: -

• 论著 • 上一篇    下一篇

采用频带能量进行ECoG信号的特征提取

赵海滨;刘冲;李春胜;王宏;   

  1. 东北大学机械工程与自动化学院;
  • 收稿日期:2013-06-20 修回日期:2013-06-20 出版日期:2010-10-15 发布日期:2013-06-20
  • 通讯作者: -
  • 作者简介:-
  • 基金资助:
    中央高校基本科研业务费专项资金资助项目(N090303002);;

Feature extraction of ECoG signals using band power

Zhao, Hai-Bin (1); Liu, Chong (1); Li, Chun-Sheng (1); Wang, Hong (1)   

  1. (1) School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110004, China
  • Received:2013-06-20 Revised:2013-06-20 Online:2010-10-15 Published:2013-06-20
  • Contact: Zhao, H.-B.
  • About author:-
  • Supported by:
    -

RichHTML

0

PDF (PC)

979

摘要： 对于典型的采用皮层脑电图的脑-机接口系统,测试者想象左手小手指或舌头运动,提出了采用频带能量的特征提取方法.首先,把频带能量作为特征,进行导联的选择,从64导联中获取特征最明显的11导进行分析;然后采用频带能量进行皮层脑电图信号的特征提取,并利用主分量分析进行降维;最后采用Fisher线性判别分析进行两类意识任务的分类.离线分析结果表明,该方法对测试数据取得了很好的分类准确率.

关键词: 皮层脑电图, 脑-机接口, 频带能量, 线性判别分析, 特征提取

Abstract: A feature extraction algorithm was proposed using band power (BP) for a typical ECoG-based brain-computer interface (BCI) system, when the subject imaged movements of either the left small finger or the tongue. In the method the BP features are taken to select the channels, and 11 channels with most distinctive features are selected from 64 channels for analysis. Then, the features of ECoG signals are extracted according to BP, and the dimensions of feature vector are reduced with principal components analysis (PCA). Consequently, two different mental tasks are classified via Fisher linear discriminant analysis (LDA). The results of off-line analysis showed that the algorithm proposed has high classification accuracy for test data set.

中图分类号: