文章摘要

文章信息/Info

作者:: 易慧1; 陈瑞娟1△; 邓光华1; 胡淑芬1; 肖淑绵1; 王慧泉1; 2; 1.天津工业大学生命科学学院，天津 300387；2.天津市光电检测技术与系统重点实验室，天津工业大学，天津 300387

Author(s):: YI Hui1; CHEN Ruijuan1; DENG Guanghua1; HU Shufen1; XIAO Shumian1; Wang Huiquan1; 2; 1.School of Life Sciences,Tianjin Polytechnic University, Tianjin 300387, China；2.Tianjin Key Laboratory of Photoelectric Testing Technology and System,Tianjin Polytechnic University,Tianjin 300387

Keywords:: Emotion recognition; Heart rate variability; Wavelet transform; Support vector machine; Electrocardiograph

摘要:: 为探究不同情绪下的自主神经系统的特征变化，本研究基于心率变异性的特征信息研究了不同情绪的分类识别。采集了恬静、气愤、悲伤、恐惧、厌恶、平静、惊奇、搞笑等八种情绪的心电信号，采用小波变换去除噪声干扰，提取了心率变异性的时域、频域等特征参数，并利用支持向量机算法实现了情绪分类。结果表明，本研究方法对于情绪识别具有较高准确性，通过优化支持向量机的参数，对于不同情绪的识别精度可达到60%~75%。基于心率变异性的特征信息进行多种情绪的分类识别方法在影片的测评、培训效果评价等方面具有重要意义。

Abstract:: In order to explore the characteristic changes of autonomic nervous system under different emotions,emotions were classified and identified based on the characteristic information of heart rate variability(HRV).We collected electrocardiograph(ECG) signals of eight emotions including quiet, anger, sadness, fear, disgust, calmness, surprise and amusement, used wavelet transform to remove noise interference, the characteristic parameters of HRV in time domain and frequency domain were extracted,and emotion classification was realized by support vector machine algorithm.The results showed that the proposed method was accurate for emotion recognition, the accuracy reached to 60%~75% by optimizing the eigenvalues and adjusting the algorithm parameters.The method to classify and identify various emotions based on the characteristic information of heart rate variability is of great significance in film evaluation and training evaluation.