DocumentCode :
568797
Title :
Emotion detection using sub-image based features through human facial expressions
Author :
Kudiri, Krishna Mohan ; Said, Abas Md ; Nayan, M. Yunus
Author_Institution :
Comput. & Inf. Sci. Dept., Univ. Teknol. PETRONAS, Tronoh, Malaysia
Volume :
1
fYear :
2012
fDate :
12-14 June 2012
Firstpage :
332
Lastpage :
335
Abstract :
The human face is an important human body part which plays an extraordinary role in the human to human or human to machine communications. As such, it is important to design robust emotion detection system for real world applications like human decision making and effective human computer interaction. Facial expression provides non-verbal communication for human computer interactions. This study identifies the problem of loss of data in the feature extraction scheme based on limited number of positions of facial muscles. To improve detection performance, relative sub-image based features are proposed. Classifications have been done using the support vector machine to implement an automated emotion detection system for facial expressions. The results show that the proposed relative sub-image based features enhance the classification rates.
Keywords :
decision making; emotion recognition; face recognition; feature extraction; human computer interaction; image classification; classification rates; detection performance; facial muscles; human computer interaction; human decision making; human facial expressions; human to human communications; human to machine communications; nonverbal communication; robust emotion detection system; subimage based features; Accuracy; Complexity theory; Image color analysis; Psychology; Support vector machines; Synchronization; Relative sub-image based features; mel-frequency cepstral coefficients; principal component analysis; support vector machine;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer & Information Science (ICCIS), 2012 International Conference on
Conference_Location :
Kuala Lumpeu
Print_ISBN :
978-1-4673-1937-9
Type :
conf
DOI :
10.1109/ICCISci.2012.6297264
Filename :
6297264
Link To Document :
بازگشت