资源描述:
《基于事件相关电位的测谎方法研究本科毕业论文》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、题目:基于事件相关电位的测谎方法研究学院 信息科学与工程学院 专业 自动化 年级 2010级 学号 姓名 指导老师 2014年5月第1章摘要本研究主要利用P300电位作为辨别测谎的重要指标。P300测谎与任务相关性、刺激性质、注意、记忆、电生理干扰等影响因素有关。利用P300测谎的目的是对单一个体进行其是否在说谎的判断,本课题研究主要关注单个被试内的探测刺激和无关刺激的P300是否存在显著差异。本文采用个人相关信息为刺激内容,以被试真实的相关信息作为目标刺激,例如被
2、试的存钱银行名称等,设计实验范式,将实验采集到的脑电信号作为主要研究对象。P300电位的数据处理分为三步:预处理,特征提取和分类。将采集到的数据通过汉明窗低通滤波器滤波,选择导联和时间窗进行降维,然后利用Fisher线性判别进行分类识别。对分类结果进行分析得出:正确率与训练量的多少、刺激重复次数、导联选取个数均有关系。关键词:事件相关电位,P300电位,Fisher线性判别ABSTRACTThisresearchmainlyusestheP300potentialasanimportantindextodis
3、tinguishthepolygraph.TherelevantfactorsofP300polygraphandtaskdependencies,stimulatingproperties,attention,memory,electrophysiologicaldisturbance.UsingP300lietothesingleindividualislyingjudgment,thisresearchmainlyfocusonthedetectionofsinglesubjectsandtndiffe
4、rentstimuliP300whethertherearesignificantdifferences.Thispaperusespersonalinformationrelatedtostimulatethecontent,relatedtotheinformationistrueastargetstimuli,suchasthebankname,designexperimentalparadigm,theEEGsignalisacquiredfromexperimentsasthemainobjecto
5、fstudy.DataprocessingofP300potentialisdividedintothreesteps:preprocessing,featureextractionandclassification.ThedatawillbecollectedbytheHammingwindowlow-passfilter,selectleadandtimewindowtoreducethedimension,andthenusetheFisherlinearclassificationdiscrimina
6、tion.Theclassificationresultsareobtained:thecorrectrateandamountoftraining,stimulusrepetition,leadthechosennumberof.Keywords:eventrelatedpotential,P300,Fisherlineardiscriminant目录第一章绪论·····················································11.1引言···············
7、················································11.2国内外ERP测谎的研究发展概况·······································11.3本文的研究内容·····················································2第二章ERP用于测谎的原理和实验范式设计··························32.1事件相关电位的概述···························
8、······················32.2P300信号介绍·······················································32.3P300测谎的影响因素················································42.4实验设计·········································