基于情感心电信号的去趋势波动分析研究  

Detrended Fluctuation Analysis Based on Emotional ECG

作　　者：程静[1] 刘光远[2] 

机构地区：[1]西南大学计算机与信息科学学院,重庆400715 [2]西南大学电子信息工程学院,重庆400715

出　　处：《西南大学学报（自然科学版）》2016年第2期169-175,共7页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金(61472330)

摘　　要：心电信号作为一种重要的生理信号,已证明其中包含可靠情感信息.在实验室诱发情感心电信号过程中,采用2遍情感视频播放机制,在第二遍观看视频过程中获取了记录被试主观情绪体验的情感重评按键文件,据此截取可靠的情感心电信号.通过比较多种去趋势波动分析算法,结果显示CMA算法的性能最为稳定.因此,采用CMA算法来计算情感心电信号的标度指数.结果显示,高兴、悲伤、愤怒和恐惧的心电信号均具有长程相关性.以标度指数作为情感特征,采用Fisher分类器进行二分类的情感识别,高兴、悲伤、愤怒和恐惧4种情感的正确识别率分别为89.56%,90.10%,70.43%,83.18%,说明情感心电信号的非线性特征对于识别目标情感具有很好的区分度.ECG（electrocardiogram）signal is one of the most important physiological signals,which has been proven to contain reliable emotional information.In a study reported herein,the emotional videos were played two times in the process of the laboratory-induced emotional ECG,and in the second presentation the press file of emotional re-evaluation was obtained,which registered the subjective experience of the participants and helped to cut the reliable emotional ECG.The detrended fluctuation analysis（DFA）was then used to get the scale indexes of emotional ECG,and the result showed that ECGs of happiness,sadness,anger and fear were long-range dependent.Finally,the scale indexes were used by the binary classifier of Fisher as the emotional features,and the result showed that the correct identification rate of happiness,sadness,anger and fear was 89.56%,90.10%,70.43% and 83.18%,respectively.In conclusion,the whole experiment displayed that the nonlinear features have fine distinction in different emotions.

关 键 词：情感心电信号 情感重评按键文件 去趋势波动分析 标度指数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]