In this paper, the authors present an experiment performed in Padua, Italy to model a cultural heritage object from multiple range views. This object is Porta Portello, an old monumental door near the University, and has been scanned using the laser scanner Cyrax 2500 from Cyra Technologies. A multi-resolution approach was carried out: 1 cm and 2 cm grids were used to acquire laser data for the front and the back sides; instead, 0.5 cm grid was used to acquire more detailed scenes of some artistic features. A typical approach is based on the closest point caching to minimize the computational resources. In order to develop an unique multi-resolution model, two registration techniques were taken in account: the point-to-point alignment and the point-plane alignment. First of all, the pair-wise registration has been performed using some high resolution range views. After this phase, a global registration was produced using all the 11.000.000 of the acquired cloud points. Some patches were also used to minimize the noise of the acquired data, when “moving” objects have been scanned together the “stable” surfaces. Furthermore, the LOD (Level Of Detail) approach was used to simplify the 3D scenes to reach a good compromise among accuracy, topology and data decimation.

Multi-resolution modelling from multiple range views: the laser scanner survey of Porta Portello, Padua, Italy

SALEMI, GIUSEPPE
;
ACHILLI, VLADIMIRO;
2004

Abstract

In this paper, the authors present an experiment performed in Padua, Italy to model a cultural heritage object from multiple range views. This object is Porta Portello, an old monumental door near the University, and has been scanned using the laser scanner Cyrax 2500 from Cyra Technologies. A multi-resolution approach was carried out: 1 cm and 2 cm grids were used to acquire laser data for the front and the back sides; instead, 0.5 cm grid was used to acquire more detailed scenes of some artistic features. A typical approach is based on the closest point caching to minimize the computational resources. In order to develop an unique multi-resolution model, two registration techniques were taken in account: the point-to-point alignment and the point-plane alignment. First of all, the pair-wise registration has been performed using some high resolution range views. After this phase, a global registration was produced using all the 11.000.000 of the acquired cloud points. Some patches were also used to minimize the noise of the acquired data, when “moving” objects have been scanned together the “stable” surfaces. Furthermore, the LOD (Level Of Detail) approach was used to simplify the 3D scenes to reach a good compromise among accuracy, topology and data decimation.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS 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/11577/2452312
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact