A Method of Evaluating the Internal Precision of Multi-View Stereo Dense Reconstruction, Applied on Parthenon Frieze
Journal
Photogrammetrie - Fernerkundung - Geoinformation
Date Issued
June 2014
DOI
10.1127/1432-8364/2014/0219
Abstract
Recent advances in Structure-fromMotion
(SfM) and multi-view stereo (MVS) techniques
are getting a lot of attention in the 3D modeling
community, as they provide a sophisticated,
versatile and rapid way to compute dense 3D point
clouds. Many applications in archeology and aerial,
surface or underwater surveys have been reported
based in the combination of SfM and MVS. In addition,
they become more popular and widespread
as commercial applications and Internet services
are emerging. Nevertheless, there is no known way
to evaluate the precision of the extracted point
cloud. This paper presents a method for estimating
the precision of each point in the final point cloud
generated by open source MVS, as it is applied to
the Parthenon frieze in the Acropolis Museum.
(SfM) and multi-view stereo (MVS) techniques
are getting a lot of attention in the 3D modeling
community, as they provide a sophisticated,
versatile and rapid way to compute dense 3D point
clouds. Many applications in archeology and aerial,
surface or underwater surveys have been reported
based in the combination of SfM and MVS. In addition,
they become more popular and widespread
as commercial applications and Internet services
are emerging. Nevertheless, there is no known way
to evaluate the precision of the extracted point
cloud. This paper presents a method for estimating
the precision of each point in the final point cloud
generated by open source MVS, as it is applied to
the Parthenon frieze in the Acropolis Museum.