摘要
针对脑-计算机接口技术中的脑电信号处理、事件相关同步和事件相关去同步的特点,提出了一种基于离散小波滤波和AR模型来提取脑电信号特征向量的方法。利用Daubechies类小波函数对脑电信号进行4层分解,然后使用Burg算法提取脑电信号8阶AR模型系数,最后用BP神经网络进行分类和比较。得到最优的正确率为71.64%,小波滤波的效果要优于FIR滤波器。
Due to the feature of event related synchronization and event related desynchronization,identification and classification technology of electroencephalography (EEG) plays an important role in the study of brain-computer interface(BCI) system. A novel method of extracting feature vector of EEG based on discrete wavelet filter and au- toregressive(AR) model was proposed. First, the EEG signal was decomposed to four levels by Daubechies wavelet function and then its eight order AR coefficients were estimated by Burg's algorithm. At last, the features were classified by BP neural network, the best accuracy is 71.64% , and the effect of wavelet filter is better than FIR filter's.
出处
《电子器件》
CAS
北大核心
2012年第4期461-464,共4页
Chinese Journal of Electron Devices
基金
江苏省普通高校研究生科研创新计划项目(CXLX12_0095)
关键词
脑电信号
小波滤波
AR模型
BP神经网络
electroencephalography
discrete wavelet filter
AR model
BP neural network