多模态情感识别综述  

Survey on Multimodal Emotion Recognition

作　　者：贾俊佳 蒋惠萍[1] 张廷[1] JIA Junjia;JIANG Huiping;ZHANG Ting(Lab.of Afective Computing,School of Information Engineering,Minzu Unirersity of China,Beijing 100081,China)

机构地区：[1]中央民族大学信息工程学院情感计算实验室

出　　处：《中央民族大学学报（自然科学版）》2020年第1期54-58,共5页Journal of Minzu University of China(Natural Sciences Edition)

摘　　要：情感是一种多元的讯息表述,而情感识别又是人机交互,感情交互中中十分极其重要的问题[1]。根据情感的差异来源,本文从脑电和表情两种模态入手,概述了脑电和表情的特征提取方法,着重说明了多模态融合技术和方法,并且列举了现阶段常见的公开多模态数据库。最后总结了仍然存在的问题和今后的研究方向。Emotion is a kind of multiple information expression, and emotion recognition is human-computer interaction. It is a very important problem in emotion interaction. Based on the two modes of EEG and expression, we introduce the related feature extraction technology, focus on the multimodal emotion fusion recognition technology, and list the commonly used open multi-modal database. Finally,the existing problems in the research of multimodal emotion recognition are briefly explained, and the future research directions are prospected.

关 键 词：情感识别 脑电 表情 多模态信息融合 综述 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]