DocumentCode :
240033
Title :
Multimodal emotion recognition (MER) system
Author :
Tang, Ke ; Yun Tie ; Yang, Tao ; Ling Guan
Author_Institution :
Dept. of Electr. & Comput. Eng., Ryerson Univ., Toronto, ON, Canada
fYear :
2014
fDate :
4-7 May 2014
Firstpage :
1
Lastpage :
6
Abstract :
Today, a number of recognition systems have been proposed widely, from audio recognition to image recognition, and from two dimensional databases to three dimensional databases; the study and research on the emotion recognition system become more important than ever before. This paper shows the new research and development of the multimodal emotion recognition system (MER). There are two main categories in this MER System, a new database and the MER fusion recognition part. The MER database and recognition system. The use Microsoft XBOX KINECT sensor, the data include 2D facial images, 3D face feature points and audio wave in a concurrent time based. In the recognition system part, it use multimodal fusion level as final classifier, include decision level fusion, feature level fusion and a new fusion level combination. The MER achieves the best overall and individual emotion recognition that represent the true emotion of human bean.
Keywords :
audio signal processing; emotion recognition; face recognition; feature extraction; image classification; image fusion; visual databases; 2D facial images; 3D face feature points; MER database; MER fusion recognition part; MER system; Microsoft XBOX KINECT sensor; audio recognition; audio wave; decision level fusion; feature level fusion; fusion level combination; image recognition; multimodal emotion recognition system; multimodal fusion level; three dimensional databases; two dimensional databases; Cameras; Databases; Emotion recognition; Face; Face recognition; Feature extraction; Three-dimensional displays;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Electrical and Computer Engineering (CCECE), 2014 IEEE 27th Canadian Conference on
Conference_Location :
Toronto, ON
ISSN :
0840-7789
Print_ISBN :
978-1-4799-3099-9
Type :
conf
DOI :
10.1109/CCECE.2014.6900993
Filename :
6900993
Link To Document :
بازگشت