TY - GEN
T1 - A facial expression recognition with automatic segmentation of face regions
AU - Hernandez-Matamoros, Andres
AU - Bonarini, Andrea
AU - Escamilla-Hernandez, Enrique
AU - Nakano-Miyatake, Mariko
AU - Perez-Meana, Hector
N1 - Publisher Copyright:
© Springer International Publishing Switzerland 2015.
PY - 2015
Y1 - 2015
N2 - This paper proposes a facial expression recognition algorithm, which automatically detects and segments the face regions of interest (ROI) such as the forehead, eyes and mouth, etc. Proposed scheme initially detects the image face and segments it in two regions: forehead/eyes and mouth. Next each of these regions is segmented into N × M blocks which are characterized using 54 Gabor functions that are correlated with each one of the N × M blocks. Next the principal component analysis (PCA) is used for dimensionality reduction. Finally, the resulting feature vectors are inserted in a proposed classifier based on clustering techniques which provides recognition results closed to those provided by the support vector machine (SVM) with much less computational complexity. The experimental results show that proposed system provides a recognition rate of about 98% when only one ROI is used. This recognition rate increases to about 99% when the feature vectors of all ROIs are concatenated. This fact allows achieving recognition rates higher than 97%, even when one of the two ROI are totally occluded.
AB - This paper proposes a facial expression recognition algorithm, which automatically detects and segments the face regions of interest (ROI) such as the forehead, eyes and mouth, etc. Proposed scheme initially detects the image face and segments it in two regions: forehead/eyes and mouth. Next each of these regions is segmented into N × M blocks which are characterized using 54 Gabor functions that are correlated with each one of the N × M blocks. Next the principal component analysis (PCA) is used for dimensionality reduction. Finally, the resulting feature vectors are inserted in a proposed classifier based on clustering techniques which provides recognition results closed to those provided by the support vector machine (SVM) with much less computational complexity. The experimental results show that proposed system provides a recognition rate of about 98% when only one ROI is used. This recognition rate increases to about 99% when the feature vectors of all ROIs are concatenated. This fact allows achieving recognition rates higher than 97%, even when one of the two ROI are totally occluded.
KW - Classifier methods
KW - Face detection
KW - Facial ROI segmentation
KW - Facial expression recognition
KW - Gabor functions
KW - PCA
UR - http://www.scopus.com/inward/record.url?scp=84945937705&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-22689-7_41
DO - 10.1007/978-3-319-22689-7_41
M3 - Contribución a la conferencia
SN - 9783319226880
T3 - Communications in Computer and Information Science
SP - 529
EP - 540
BT - Intelligent Software Methodologies, Tools and Techniques - 14th International Conference, SoMeT 2015, Proceedings
A2 - Fujita, Hamido
A2 - Guizzi, Guido
PB - Springer Verlag
T2 - 14th International Conference on New Trends in Intelligent Software Methodology, Tools, and Techniques, SoMeT 2015
Y2 - 15 September 2015 through 17 September 2015
ER -