DocumentCode :
3320584
Title :
Towards generic fitting using multiple features Discriminative Active Appearance Models
Author :
Martins, Pedro ; Batista, Jorge
Author_Institution :
Dept. of Electr. Eng. & Comput., Univ. of Coimbra, Coimbra, Portugal
fYear :
2010
fDate :
26-29 Sept. 2010
Firstpage :
4545
Lastpage :
4548
Abstract :
A solution for Discriminative Active Appearance Models is proposed. The model consists in a set of descriptors which are covariances of multiple features evaluated over the neighborhood of the landmarks whose locations are governed by a Point Distribution Model (PDM). The covariance matrices are a special set of tensors that lie on a Riemannian manifold, which make it possible to measure the dissimilarity and to update them, imposing the temporal appearance consistency. The discriminative fitting method produce patch response maps found by convolution around the current landmark position. Since the minimum of the responce map isn´t always the correct solution due to detection ambiguities, our method finds candidates to solutions based on a mean-shift algorithm, followed by an unsupervised clustering technique used to locate and group the candidates. A mahalanobis based metric is used to select the best solution that is consistent with the PDM. Finally the global PDM optimization step is performed using a weighted least-squares warp update, based on the Lucas Kanade framework. The weights were extracted from a landmark matching score statistics. The effectiveness of the proposed approach was evaluated on unseen data on the challenging Talking Face video sequence, demonstrating the improvement in performance.
Keywords :
covariance matrices; curve fitting; face recognition; image sequences; least squares approximations; pattern clustering; shape recognition; tensors; video signal processing; Lucas Kanade framework; Mahalanobis based metric; Riemannian manifold; covariance matrix; descriptor; discriminative active appearance model; discriminative fitting; generic fitting; global PDM optimization; landmark position; mean-shift algorithm; patch response map; point distribution model; talking face video sequence; temporal appearance consistency; tensor; unsupervised clustering; weighted least-squares warp update; Computational modeling; Covariance matrix; Face; Manifolds; Optimization; Shape; Training; Discriminative Active Appearance Models; Point Distribution Model; Riemannian Manifolds;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Image Processing (ICIP), 2010 17th IEEE International Conference on
Conference_Location :
Hong Kong
ISSN :
1522-4880
Print_ISBN :
978-1-4244-7992-4
Electronic_ISBN :
1522-4880
Type :
conf
DOI :
10.1109/ICIP.2010.5650726
Filename :
5650726
Link To Document :
بازگشت