电子学报 ›› 2017, Vol. 45 ›› Issue (8): 1836-1841.DOI: 10.3969/j.issn.0372-2112.2017.08.005

• 学术论文 • 上一篇    下一篇

基于脑电样本熵的测谎分析

高军峰1,2, 司慧芳1, 余彬1, 顾凌云1, 梁莹1, 杨勇3   

  1. 1. 中南民族大学生物医学工程学院&认知科学国家民委重点实验室& 中南民族大学医学信息分析及肿瘤诊疗湖北省重点实验室 湖北武汉 430074;
    2. 电子科技大学生命科学与技术学院 四川成都 610054;
    3. 江西财经大学信息管理学院 江西南昌 330000
  • 收稿日期:2016-03-29 修回日期:2016-08-16 出版日期:2017-08-25
    • 通讯作者:
    • 杨勇
    • 作者简介:
    • 高军峰,男,1973年生于湖北武汉.副教授,硕士生导师,2005年和2011年分别在武汉理工大学和西安交通大学获得工学硕士和工学博士学位.现为中南民族大学教师,主要从事生物医学信号处理、神经网络和机器学习等方面的研究工作.E-mail:junfengmst@163.com;司慧芳,女,1993年出生于河南济源,硕士研究生,研究方向为脑电信号处理与模式识别.
    • 基金资助:
    • 国家自然科学基金 (No.81271659,61262034); 中国博士后科学基金 (No.2014M552346)

Lie Detection Analysis Based on the Sample Entropy of EEG

GAO Jun-feng1,2, SI Hui-fang1, YU Bin1, GU Ling-yun1, LIANG Ying1, YANG Yong3   

  1. 1. School of Biomedical Engineering, South-Central University for Nationalities & Key Laboratory of Cognitive Science, State Ethnic Affairs Commission & Hubei Key Laboratory of Medical Information Analysis and Tumor Diagnosis & Treatment, Wuhan, Hubei 430074, China;
    2. School of Life Science and Technology, Chengdu, Sichuan 610054, China;
    3. School of Information Technology, Jiangxi University of Finance and Economics, Nanchang, Jiangxi 330000, China
  • Received:2016-03-29 Revised:2016-08-16 Online:2017-08-25 Published:2017-08-25

摘要: 测谎分析在刑讯侦查和法律审判中具有重要意义.为了区分是否说谎,根据脑电信号的非线性特征,本文首次使用非线性动力学的样本熵方法分析30名受试者处于诚实和说谎两种状态时脑电信号的样本熵值.研究发现:受试者处于诚实状态时的熵值波动范围明显小于说谎状态下的波动范围,更重要的是说谎时的熵值显著高于说实话时的熵值,表明样本熵可以区分诚实和说谎两种不同状态下的脑电信号,该研究为基于脑电的测谎提供了一种新的途径.

关键词: 测谎, 脑电信号, 非线性特征, 样本熵

Abstract: There is great significance in lie detection for the criminal investigations and law trials.In this study,according to the nonlinear characteristics of electroencephalography (EEG),it is the first time to use the sample entropy (SE),a nonlinear dynamical parameter of EEG,to see if someone is lying.The sample entropy values of 30 subjects' EEG signals in lying or honesty states were calculated and analyzed.The study found that the fluctuating range of SE values in honesty was obviously less than that in lying.It is more important that the SE values in lying was significantly higher than the honesty,which indicated that SE could be used to distinguish EEG signals between two different states of honesty and lying.This research provides a new way for EEG-based lie detection.

Key words: lie detection, electroencephalography(EEG) signals, nonlinear characteristic, sample entropy

中图分类号: