重庆理工大学学报2025，Vol.39Issue(9)：82-89,8.DOI:10.3969/j.issn.1674-8425(z).2025.05.010
融合时序动态特征的fNIRS情绪识别方法研究
Research on emotion recognition method based on fnirs with integrated temporal dynamic features
摘要
Abstract
To address the inadequate capture of dynamic temporal features in functional Near-Infrared Spectroscopy(fNIRS)technology for emotion recognition,this paper proposes an emotion recognition model that integrates Temporal Dynamic Emotional Recognition Model(TDERM).By combining Long Short-Term Memory networks(LSTM)and Convolutional Neural Networks(CNN),the model employs Multi-Head Self-Attention mechanisms(MSA)along with Global Temporal Mean processing,significantly enhancing the perception of emotional state changes.Experimental results indicate TDERM accurately recognizes positive,neutral,and negative emotions in a self-constructed fNIRS emotion dataset,achieving an average recognition accuracy of over 95%under single hemoglobin concentration.Furthermore,comparative analysis of different hemoglobin concentration combinations reveals the combination of HbO2,HbR,and HbT along the time dimension(HbSum＿T)performs best across various classification models,with an average accuracy of 88.99%.This study may provide a new avenue for fNIRS technology in emotion recognition.关键词
功能性近红外光谱技术/情绪识别/时序动态特征/血红蛋白组合方式Key words
fNIRS/emotion recognition/temporal dynamic features/hemoglobin concentration combina-tion分类
信息技术与安全科学引用本文复制引用
李修军,葛雄心,李奇,武岩,杨菁菁,孟天宇..融合时序动态特征的fNIRS情绪识别方法研究[J].重庆理工大学学报,2025,39(9):82-89,8.基金项目
吉林省教育厅科学技术研究项目(JJKH20220780KJ,JJKH20230847KJ) （JJKH20220780KJ,JJKH20230847KJ）
