International Journal of Computer Applications
Foundation of Computer Science (FCS), NY, USA
|
Volume 55 - Issue 17 |
Published: October 2012 |
Authors: G. Nirmala Priya, R. S. D. Wahida Banu |
![]() |
G. Nirmala Priya, R. S. D. Wahida Banu . Person Independent Facial Expression Detection using MBWM and Multiclass SVM. International Journal of Computer Applications. 55, 17 (October 2012), 52-58. DOI=10.5120/8851-3180
@article{ 10.5120/8851-3180, author = { G. Nirmala Priya,R. S. D. Wahida Banu }, title = { Person Independent Facial Expression Detection using MBWM and Multiclass SVM }, journal = { International Journal of Computer Applications }, year = { 2012 }, volume = { 55 }, number = { 17 }, pages = { 52-58 }, doi = { 10.5120/8851-3180 }, publisher = { Foundation of Computer Science (FCS), NY, USA } }
%0 Journal Article %D 2012 %A G. Nirmala Priya %A R. S. D. Wahida Banu %T Person Independent Facial Expression Detection using MBWM and Multiclass SVM%T %J International Journal of Computer Applications %V 55 %N 17 %P 52-58 %R 10.5120/8851-3180 %I Foundation of Computer Science (FCS), NY, USA
Facial expression analysis is an attractive, challenging and important field of study in facial analysis. It's important applications include many areas such as human–computer interaction, human emotion analysis, biometric authentication, exhaustion detection and data-driven animation. For successful facial expression recognition, the first step is to arrive at an appropriate facial representation from original face image which is a crucial step. This paper, empirically evaluate facial representation using statistical features from the Local Binary Patterns, Simplified local binary mean and Mean based weight matrix for person-independent facial expression recognition. Multiclass SVM is applied systematically for classification. The Japanese female database JAFFE is used for the experiment. Extensive experiments shows that statistical features derived from LBP are effective and efficient for facial expression recognition. Further improved and best results are obtained with SLBM and MBWM features extracted using Multiclass Support Vector Machine classifiers.