论文部分内容阅读
脑机接口(BCI)是在人或动物脑与外部设备间建立的直接连接通路,信号分析功能模块是其核心部分,其中特征提取算法的效果如何是脑电图(EEG)信号分析算法的关键。EEG信号本身信噪比低,传统的EEG特征提取方法存在着缺少空间信息,需要的特征量个数较多,分类正确率低等不足。针对以上问题,本文提出了一种基于小波和独立分量分析(ICA)的时间-频率-空间EEG特征的提取方法,分别用离散小波变换(DWT)和ICA提取时频域特征和空域特征。并用支持向量机(SVM)和遗传算法(GA)相结合的方法对提取的特征进行分类。实验对比结果表明,所提出的方法有效地克服了传统的时频特征提取方法空间信息描述不足等问题,对于2003年BCI竞赛数据datasetⅢ分析,最高分类正确率为90.71%。
BCI is a direct connection between the human or animal brain and external devices. The signal analysis function module is the core part of the BCI. The effect of the feature extraction algorithm is the key of EEG signal analysis algorithm . The signal-to-noise ratio of the EEG signal itself is low. The traditional EEG feature extraction method lacks spatial information, requires a large number of feature quantities and has low classification accuracy. In order to solve the above problems, this paper presents a time-frequency-space EEG feature extraction method based on wavelet and independent component analysis (ICA). The discrete wavelet transform (DWT) and ICA are used to extract time-frequency domain features and spatial domain features respectively. The extracted features were classified by the combination of Support Vector Machine (SVM) and Genetic Algorithm (GA). Experimental results show that the proposed method can effectively overcome the problem of lack of description of spatial information in the traditional time-frequency feature extraction methods. For the dataset III analysis of the 2003 BCI competition data, the highest classification accuracy is 90.71%.