On affine invariant descriptors related to SIFT

Using a classical result on algebraic invariants of the unimodular group, we present in this paper some basic geometric affine invariant quantities, and we use them to construct some distinctive descriptors for object detection. Although full affine invariance cannot be guaranteed due to noncommutat...

Full description

Bibliographic Details
Authors: Sadek, Rida, Constantinopoulos, Constantinos, Meinhardt Llopis, Enric, Ballester, Coloma, Caselles, Vicente
Format: article
Status:Published version
Publication Date:2012
Country:España
Institution:Universitat Pompeu Fabra
Repository:Repositorio Digital de la UPF
OAI Identifier:oai:repositori.upf.edu:10230/46739
Online Access:http://hdl.handle.net/10230/46739
http://dx.doi.org/10.1137/100798739
Access Level:Open access
Keyword:Image matching
Affine invariance
Image descriptors
Object recognition
id ES_669efb927a3a3d87b52aca2f40df0b21
oai_identifier_str oai:repositori.upf.edu:10230/46739
network_acronym_str ES
network_name_str España
repository_id_str
spelling On affine invariant descriptors related to SIFTSadek, RidaConstantinopoulos, ConstantinosMeinhardt Llopis, EnricBallester, ColomaCaselles, VicenteImage matchingAffine invarianceImage descriptorsObject recognitionUsing a classical result on algebraic invariants of the unimodular group, we present in this paper some basic geometric affine invariant quantities, and we use them to construct some distinctive descriptors for object detection. Although full affine invariance cannot be guaranteed due to noncommutativity of camera blur with affine maps and the domain problem (that is, the difficulty of finding an affine covariant domain), the proposed descriptors behave more robustly than SIFT with respect to affine deformations. This is supported by our comparisons both with the version of SIFT computed on an affine normalized neighborhood, and with ASIFT, which solves both the previously mentioned camera blur and domain problems by cleverly sampling the orbit of affine transformations of the images.The work of these authors was partially supported by MICINN project, reference MTM2009-08171, and by GRC, reference 2009 SGR 773, funded by the Generalitat de Catalunya. The last author’s work was also partially supported by “ICREA Acad`emia” prize for excellence in research funded by the Generalitat de Catalunya.SIAM (Society for Industrial and Applied Mathematics)202120212012info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfapplication/pdfhttp://hdl.handle.net/10230/46739http://dx.doi.org/10.1137/100798739reponame:Repositorio Digital de la UPFinstname:Universitat Pompeu FabraInglésSIAM Journal on Imaging Sciences. 2012 Jun 5;5(2):652-87info:eu-repo/grantAgreement/ES/3PN/MTM2009-08171Copyright © by SIAM. Unauthorized reproduction of this article is prohibited. First Published in Journal on Imaging Sciences in volume 5 and number 2, year 2012, published by the Society for Industrial and Applied Mathematics (SIAM).info:eu-repo/semantics/openAccessoai:repositori.upf.edu:10230/467392026-06-12T07:21:37Z
dc.title.none.fl_str_mv On affine invariant descriptors related to SIFT
title On affine invariant descriptors related to SIFT
spellingShingle On affine invariant descriptors related to SIFT
Sadek, Rida
Image matching
Affine invariance
Image descriptors
Object recognition
title_short On affine invariant descriptors related to SIFT
title_full On affine invariant descriptors related to SIFT
title_fullStr On affine invariant descriptors related to SIFT
title_full_unstemmed On affine invariant descriptors related to SIFT
title_sort On affine invariant descriptors related to SIFT
dc.creator.none.fl_str_mv Sadek, Rida
Constantinopoulos, Constantinos
Meinhardt Llopis, Enric
Ballester, Coloma
Caselles, Vicente
author Sadek, Rida
author_facet Sadek, Rida
Constantinopoulos, Constantinos
Meinhardt Llopis, Enric
Ballester, Coloma
Caselles, Vicente
author_role author
author2 Constantinopoulos, Constantinos
Meinhardt Llopis, Enric
Ballester, Coloma
Caselles, Vicente
author2_role author
author
author
author
dc.subject.none.fl_str_mv Image matching
Affine invariance
Image descriptors
Object recognition
topic Image matching
Affine invariance
Image descriptors
Object recognition
description Using a classical result on algebraic invariants of the unimodular group, we present in this paper some basic geometric affine invariant quantities, and we use them to construct some distinctive descriptors for object detection. Although full affine invariance cannot be guaranteed due to noncommutativity of camera blur with affine maps and the domain problem (that is, the difficulty of finding an affine covariant domain), the proposed descriptors behave more robustly than SIFT with respect to affine deformations. This is supported by our comparisons both with the version of SIFT computed on an affine normalized neighborhood, and with ASIFT, which solves both the previously mentioned camera blur and domain problems by cleverly sampling the orbit of affine transformations of the images.
publishDate 2012
dc.date.none.fl_str_mv 2012
2021
2021
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/10230/46739
http://dx.doi.org/10.1137/100798739
url http://hdl.handle.net/10230/46739
http://dx.doi.org/10.1137/100798739
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv SIAM Journal on Imaging Sciences. 2012 Jun 5;5(2):652-87
info:eu-repo/grantAgreement/ES/3PN/MTM2009-08171
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv SIAM (Society for Industrial and Applied Mathematics)
publisher.none.fl_str_mv SIAM (Society for Industrial and Applied Mathematics)
dc.source.none.fl_str_mv reponame:Repositorio Digital de la UPF
instname:Universitat Pompeu Fabra
instname_str Universitat Pompeu Fabra
reponame_str Repositorio Digital de la UPF
collection Repositorio Digital de la UPF
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869409829493145600
score 15,811543

Similar Items