生理信号的数据采集及其在情绪识别中的应用

发布时间：2019-04-30 06:47

【摘要】：个体长期处于负面情绪下会导致情绪系统紊乱,并且对环境适应能力及日常学习都会造成影响,然而这类个体的情绪多不易被外界察觉。多生理信号情绪识别是通过分析多生理信号的特征参量或特征参量的组合监测人体生理指标的变化,反馈出个体真实的情绪活动状态。心理干预是通过心理学的理论和方法来处理问题,让人们失衡的认知和情绪状态趋于稳定。本文运用混沌理论对个体的多生理信号进行情绪识别,并在验证此方法可行性基础上,对识别结果为负面情绪的个体通过心理干预进行心理辅导。论文首先采用生理信号仪器提取两名志愿者(一男一女)四种不同的情绪(悲伤、愉快、愤怒、高兴)下的多生理信号(心电信号、呼吸信号,皮电信号)的混沌特征参量(最大李亚普诺夫指数、信息熵、近似熵、盒子维和复杂度)。其次将提取出的混沌特征参量组成混沌特征矩阵,结合C5.0决策树分类器对4种情绪识别并分类。实验结果表明,C5.0决策树在基于混沌理论多特征参量进行情绪识别方面识别率分别为91%和93%,且个体性别差异影响不大。在此基础上追踪一名识别结果为负面情绪的个体,通过认知行为治疗法,利用生理信号仪器找到引起个体负面情绪的热点,采用PR-II生物反馈放松训练仪和心理干预共同对个体进行心理辅导。研究表明,被试者正面情绪的识别率由60%上升到100%,说明基于情绪识别的心理干预有较好的情绪干预效果。
[Abstract]:Being in negative mood for a long time can lead to disorder of emotional system and affect environmental adaptability and daily learning. However, the emotion of this kind of individual is difficult to be detected by the outside world. Multi-physiological signal emotion recognition is to monitor the changes of physiological indexes by analyzing the characteristic parameters or the combination of characteristic parameters of multi-physiological signals, so as to feedback the real emotional state of the individual. Psychological intervention is to deal with problems through psychological theory and methods, so that people's unbalanced cognitive and emotional state tends to stabilize. In this paper, we use chaos theory to recognize individual's multi-physiological signals, and on the basis of verifying the feasibility of this method, we give psychological counseling to individuals whose negative emotion is identified by psychological intervention. Firstly, physiological signal instruments were used to extract multiple physiological signals (ECG, respiratory signals) from two volunteers (a man and a woman) under four different emotions (sadness, pleasure, anger, happiness). Chaos characteristic parameters (maximum Lyapunov exponent, information entropy, approximate entropy, box and complexity). Secondly, chaotic feature matrix is composed of the extracted chaotic characteristic parameters, and four kinds of emotions are identified and classified by using C5.0 decision tree classifier. The experimental results show that the recognition rate of C5.0 decision tree is 91% and 93% respectively for emotion recognition based on multi-feature parameters of chaos theory, and the gender difference is not significant. On this basis, a person who identified the negative emotion as a result was tracked, and the hot spots that caused the negative emotion were found by means of cognitive behavioral therapy and physiological signal instrument. PR-II biofeedback relaxation training instrument and psychological intervention were used to provide psychological guidance to individuals. The results showed that the recognition rate of positive emotion increased from 60% to 100%, which indicated that psychological intervention based on emotion recognition had a better effect of emotional intervention.
【学位授予单位】：长春大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318;TP274.2

【参考文献】