Consistent Pose Normalization of Non-Rigid Shapes using One-Class Support Vector Machines

Abstract : The estimation of 3D surface correspondence constitutes a fundamental problem in shape matching and analysis ap- plications. In the presence of non-rigid shape deformations, the ambiguity of surface correspondence increases together with the complexity of registration algorithms. In this pa- per, we alleviate this problem by using One-Class Support Vector Machines (OCSVM) in order to normalize the pose of 3D objects. We show how OCSVM are employed in order to increase the consistency of translation and scale normal- ization under articulations, extrusions or the presence of outliers. To estimate the relative translation and scale of an object, we use the 3D distribution of points that is mod- elled by employing OCSVM to estimate the decision surface corresponding to the surface points of the object. To evalu- ate the performance, we use a dataset of 3D objects where we introduce various extrusions, articulations or outliers and demonstrate the increased robustness of the proposed methodology.
Type de document :
Communication dans un congrès
Computer Vision and Pattern Recognition Workshops, 2011, Colorado, United States. 2011, 〈10.1109/CVPRW.2011.5981714〉
Liste complète des métadonnées

Littérature citée [21 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-00758993
Contributeur : Panagiotis Papadakis <>
Soumis le : jeudi 29 novembre 2012 - 16:51:11
Dernière modification le : mercredi 26 septembre 2018 - 16:32:03
Document(s) archivé(s) le : samedi 17 décembre 2016 - 17:37:41

Fichier

papadakis_nordia2011.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Citation

Panagiotis Papadakis, Fiora Pirri. Consistent Pose Normalization of Non-Rigid Shapes using One-Class Support Vector Machines. Computer Vision and Pattern Recognition Workshops, 2011, Colorado, United States. 2011, 〈10.1109/CVPRW.2011.5981714〉. 〈hal-00758993〉

Partager

Métriques

Consultations de la notice

68

Téléchargements de fichiers

87