K-Means Clustering-Based Kernel Canonical Correlation Analysis for Multimodal Emotion Recognition in Human-Robot Interaction

被引：62

作者：

Chen, Luefeng ^{[1
,2
,3
]}

Wang, Kuanlin ^{[1
,2
,3
]}

Li, Min ^{[1
,2
,3
]}

Wu, Min ^{[1
,2
,3
]}

Pedrycz, Witold ^{[4
,5
,6
]}

Hirota, Kaoru ^{[7
]}

机构：

[1] China Univ Geosci, Sch Automat, Wuhan 430074, Peoples R China

[2] Hubei Key Lab Adv Control & Intelligent Automat C, Wuhan 430074, Peoples R China

[3] Minist Educ, Engn Res Ctr Intelligent Technol Geoexplorat, Wuhan 430074, Peoples R China

[4] Univ Alberta, Dept Elect & Comp Engn, Edmonton, AB T6R 2V4, Canada

[5] King Abgudulaziz Univ, Fac Engn, Dept Elect & Comp Engn, Jeddah 21589, Saudi Arabia

[6] Polish Acad Sci, Syst Res Inst, PL-01447 Warsaw, Poland

[7] Tokyo Inst Technol, Yokohama, Kanagawa 2268502, Japan

来源：

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS | 2023年 / 70卷 / 01期

基金：

中国国家自然科学基金;

关键词：

Feature fusion; K-means clustering; Kernel canonical correlation analysis (KCCA); multimodal emotion recognition; REGRESSION; FEATURES; PATTERN;

D O I：

10.1109/TIE.2022.3150097

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

In this article, K-means clustering-based Kernel canonical correlation analysis algorithm is proposed for multimodal emotion recognition in human-robot interaction (HRI). The multimodal features (gray pixels; time and frequency domain) extracted from facial expression and speech are fused based on Kernel canonical correlation analysis. K-means clustering is used to select features from multiple modalities and reduce dimensionality. The proposed approach can improve the heterogenicity among different modalities and make multiple modalities complementary to promote multimodal emotion recognition. Experiments on two datasets, namely SAVEE and eNTER-FACE'05, are conducted to evaluate the accuracy of the proposed method. The results show that the proposed method produces good recognition rates that are higher than the ones produced by the methods without K-means clustering; more specifically, they are 2.77% higher in SAVEE and 4.7% higher in eNTERFACE'05.

引用

页码：1016 / 1024

页数：9

共 32 条

[1] Facial Expression Recognition Using Local Transitional Pattern on Gabor Filtered Facial Images [J].