基于ECG和HRV特征提取的情绪状态分析

Analysis of Emotional States Based on ECG and HRV Feature Extraction

心电(ECG)和心率变异性(HRV)是心脏活动的直接响应,与自主神经系统有着密切关系,不同情绪状态有特定的自主神经反应模式,因而ECG和HRV可以反映人体情绪变化.本文利用Augsburg大学建立的情感生理数据库,选取音乐情感诱发下的心电信号,对其进行滤波处理去除噪声和干扰;采用小波变换提取心电信号小波系数的模极值和过零点,得到心电R波信号,计算其一阶差分,获得4种情绪状态下的心率变异性特征信号;提取情绪诱发的心电和心率变异性信号的时域、频域和小波熵时频特征,并对特征进行差异统计检验和情绪状态分析.仿真结果表明,HRV特征相比ECG特征可以更有效地反映人体的情绪状态,小波熵特征和HRV频谱总功率特征对于不同情绪状态的差异性更显著,该研究可为机器情绪识别提供借鉴.

Electrocardiogram(ECG)and heart rate variability(HRV)can directly response to cardiac activity,and are closely related to the autonomic nervous system.The different emotional states have specific autonomic nervous response patterns.Therefore,ECG and HRV can reflect human emotion changes.This paper applies,apply the emotional physiological database established by Augsburg University,selectS the ECG signals induced by different emotional music,and filter various noises and interferences from ECG signal.the wavelet transform is used to extract the modulus extreme value and zero-crossing point of the wavelet coefficients,detect R wave peak of ECG signals,calculate the first-order difference,and obtain the characteristic signal of heart rate variability under four emotion states.Finally,extract the time-domain,frequency-domain and wavelet entropy time-frequency features for ECG and HRV signals,which were evoked by the different emotions,,as well as perform statistical test and emotional analysis on the features.The simulation results show that HRV features can more effectively reflect the emotional states than ECG features.,the feature differences of both wavelet entropy and HRV spectrum total power are more significant for different emotional states.This research can provide the reference for machine emotion recognition.