Finding the most relevant symmetry planes for an object is a key step in many computer vision and object recognition tasks. In fact such information can be effectively used as a starting point for object segmentation, noise reduction, alignment and recognition. Some of these applications are strongly affected by the accuracy of symmetry planes estimation, thus the use of a technique that is both accurate and robust to noise is critical. In this paper we introduce a new weighted association graph which relates the main symmetry planes of 3D objects to large sets of tightly coupled vertices. This technique allows us to cast symmetry detection to a classical pairwise clustering problem, which we solve using the very effective Dominant Sets framework. The improvement of our approach over other well known techniques is shown with several tests over both synthetic data and sampled point clouds.

Consensus Graphs for Symmetry Plane Estimation

ALBARELLI, Andrea;PELILLO, Marcello;
2008-01-01

Abstract

Finding the most relevant symmetry planes for an object is a key step in many computer vision and object recognition tasks. In fact such information can be effectively used as a starting point for object segmentation, noise reduction, alignment and recognition. Some of these applications are strongly affected by the accuracy of symmetry planes estimation, thus the use of a technique that is both accurate and robust to noise is critical. In this paper we introduce a new weighted association graph which relates the main symmetry planes of 3D objects to large sets of tightly coupled vertices. This technique allows us to cast symmetry detection to a classical pairwise clustering problem, which we solve using the very effective Dominant Sets framework. The improvement of our approach over other well known techniques is shown with several tests over both synthetic data and sampled point clouds.
Structural, Syntactic, and Statistical Pattern Recognition. IAPR International Workshop on
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in ARCA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10278/31137
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? 0
social impact