Probabilistic Deformable Surface Tracking From Multiple Videos

Cedric Cagniart; Edmond Boyer; Slobodan Ilic

doi:10.1007/978-3-642-15561-1_24

Probabilistic Deformable Surface Tracking From Multiple Videos

Cedric Cagniart ¹ Edmond Boyer ² Slobodan Ilic ¹

1 CAMPAR - Computer Aided Medical Procedures & Augmented Reality

2 PERCEPTION - Interpretation and Modelling of Images and Videos

Inria Grenoble - Rhône-Alpes, Institut National Polytechnique de Grenoble (INPG), LJK - Laboratoire Jean Kuntzmann

Abstract : In this paper, we address the problem of tracking the temporal evolution of arbitrary shapes observed in multi-camera setups. This is motivated by the ever growing number of applications that require consistent shape information along temporal sequences. The approach we propose considers a temporal sequence of independently reconstructed surfaces and iteratively deforms a reference mesh to fit these observations. To effectively cope with outlying and missing geometry, we introduce a novel probabilistic mesh deformation framework. Using generic local rigidity priors and accounting for the uncertainty in the data acquisition process, this framework effectively handles missing data, relatively large reconstruction artefacts and multiple objects. Extensive experiments demonstrate the effectiveness and robustness of the method on various 4D datasets.

Type de document :

Communication dans un congrès

Kostas Daniilidis and Petros Maragos and Nikos Paragios. ECCV 2010 - 11th European Conference on Computer Vision, Sep 2010, Heraklion, Greece. Springer, 6314, pp.326-339, 2010, Lecture Notes in Computer Science; Computer Vision - ECCV 2010. <10.1007/978-3-642-15561-1_24>

Domaine :

Informatique [cs] / Vision par ordinateur et reconnaissance de formes [cs.CV]

https://hal.inria.fr/inria-00568912
Contributeur : Edmond Boyer <>
Soumis le : mercredi 23 février 2011 - 19:59:56
Dernière modification le : mardi 15 juillet 2014 - 22:03:55

Probabilistic Deformable Surface Tracking From Multiple Videos

Fichiers

Identifiants

Collections

Citation

Exporter

Partager

Métriques

348

375

Contact