TY - JOUR
T1 - Active shape models with invariant optimal features
T2 - Application to facial analysis
AU - Sukno, Federico M.
AU - Ordás, Sebastián
AU - Butakoff, Constantine
AU - Cruz, Santiago
AU - Frangi, Alejandro F.
N1 - Funding Information:
This work was partially funded by grants TIC2002-04495-C02 and TEC2006-03617/TCM, from the Spanish Ministry of Education & Science, and grant FIT-360000-2006-55 from the Spanish Ministry of Industry. Federico M. Sukno is supported by a BSCH grant. Sebastián Ordás is supported by an FPU grant from the Spanish Ministry of Education & Science. Alejandro F. Frangi holds a Ramón y Cajal Research Fellowship. The Computational Imaging Lab at Pompeu Fabra University is a member of the Biosecure (IST-2002-507534) European Network of Excellence.
PY - 2007/7
Y1 - 2007/7
N2 - This work is framed in the field of statistical face analysis. In particular, the problem of accurate segmentation of prominent features of the face in frontal view images is addressed. We propose a method that generalizes linear Active Shape Models (ASMs), which have already been used for this task. The technique is built upon the development of a nonlinear intensity model, incorporating a reduced set of differential invariant features as local image descriptors. These features are invariant to rigid transformations, and a subset of them is chosen by Sequential Feature Selection for each landmark and resolution level. The new approach overcomes the unimodality and Gaussianity assumptions of classical ASMs regarding the distribution of the intensity values across the training set. Our methodology has demonstrated a significant improvement in segmentation precision as compared to the linear ASM and Optimal Features ASM (a nonlinear extension of the pioneer algorithm) in the tests performed on AR, XM2VTS, and EQUINOX databases.
AB - This work is framed in the field of statistical face analysis. In particular, the problem of accurate segmentation of prominent features of the face in frontal view images is addressed. We propose a method that generalizes linear Active Shape Models (ASMs), which have already been used for this task. The technique is built upon the development of a nonlinear intensity model, incorporating a reduced set of differential invariant features as local image descriptors. These features are invariant to rigid transformations, and a subset of them is chosen by Sequential Feature Selection for each landmark and resolution level. The new approach overcomes the unimodality and Gaussianity assumptions of classical ASMs regarding the distribution of the intensity values across the training set. Our methodology has demonstrated a significant improvement in segmentation precision as compared to the linear ASM and Optimal Features ASM (a nonlinear extension of the pioneer algorithm) in the tests performed on AR, XM2VTS, and EQUINOX databases.
KW - Face and gesture recognition
KW - Feature evaluation and selection
KW - Invariants
KW - Shape model
KW - Statistical image analysis
UR - https://www.scopus.com/pages/publications/34249683911
U2 - 10.1109/TPAMI.2007.1041
DO - 10.1109/TPAMI.2007.1041
M3 - Article
C2 - 17496371
SN - 0162-8828
VL - 29
SP - 1105
EP - 1117
JO - IEEE Transactions on Pattern Analysis and Machine Intelligence
JF - IEEE Transactions on Pattern Analysis and Machine Intelligence
IS - 7
ER -