基于AR模型的脑电信号身份识别研究  

Study on the Identity Recognition of EEG Signals Based on the AR Model

作　　者：穆振东[1] 王平[1] 

机构地区：[1]江西蓝天学院信息研究所

出　　处：《江西蓝天学院学报》2010年第2期50-54,共5页Journal of Jiangxi Blue Sky University

基　　金：江西省教育厅2008年度科技计划项目“基于Rough集的脑电信号特征提取和分类方法研究”(GJJ08477)阶段性成果

摘　　要：本文提出利用脑电信号作为身份识别工具,利用AR模型把时域上的脑电信号转换到频域上,计算Fisher距离进行特征提取,分析特征在频率段的分布,得出特征主要分布在14～30Hz之间.利用BP神经网络进行计算,身份平均识别率达到86.7%。The paper puts forward that by taking EEG signals as the identity recognition tool and using the AR model to transfer EEG signals from time domain to frequency domain,we can calculate Fisher distance to abstract features and analyze the distribution of features in frequency section and get the conclusion that the features are distributed mainly between 14 and 30 Hz. Applying BP neutral network to calculate,the average recognition rate of identity reaches up to 86.7%.

关 键 词：脑电信号 AR模型 身份识别 BP神经网络 

分 类 号：TP802.4[自动化与计算机技术—检测技术与自动化装置]