A scalable, efficient, and accurate solution to non-rigid structure from motion

Most Non-Rigid Structure from Motion (NRSfM) solutions are based on factorization approaches that allow reconstructing objects parameterized by a sparse set of 3D points. These solutions, however, are low resolution and generally, they do not scale well to more than a few tens of points. While there...

Descripción completa

Detalles Bibliográficos
Autores: Agudo Martínez, Antonio, Moreno-Noguer, Francesc
Tipo de recurso: artículo
Estado:Versión aceptada para publicación
Fecha de publicación:2018
País:España
Institución:Consejo Superior de Investigaciones Científicas (CSIC)
Repositorio:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/167144
Acceso en línea:http://hdl.handle.net/10261/167144
Access Level:acceso abierto
Palabra clave:Factorization
Non-rigid structure from motion
Low-rank representation
Time-varying scenes
Probabilistic trajectory space
id ES_9cca115856b2494275a0d2cae202cdbd
oai_identifier_str oai:digital.csic.es:10261/167144
network_acronym_str ES
network_name_str España
repository_id_str
spelling A scalable, efficient, and accurate solution to non-rigid structure from motionAgudo Martínez, AntonioMoreno-Noguer, FrancescFactorizationNon-rigid structure from motionLow-rank representationTime-varying scenesProbabilistic trajectory spaceMost Non-Rigid Structure from Motion (NRSfM) solutions are based on factorization approaches that allow reconstructing objects parameterized by a sparse set of 3D points. These solutions, however, are low resolution and generally, they do not scale well to more than a few tens of points. While there have been recent attempts at bringing NRSfM to a dense domain, using for instance variational formulations, these are computationally demanding alternatives which require certain spatial continuity of the data, preventing their use for articulated shapes with large deformations or situations with multiple discontinuous objects. In this paper, we propose incorporating existing point trajectory low-rank models into a probabilistic framework for matrix normal distributions. With this formalism, we can then simultaneously learn shape and pose parameters using expectation maximization, and easily exploit additional priors such as known point correlations. While similar frameworks have been used before to model distributions over shapes, here we show that formulating the problem in terms of distributions over trajectories brings remarkable improvements, especially in generality and efficiency. We evaluate the proposed approach in a variety of scenarios including one or multiple objects, sparse or dense reconstructions, missing observations, mild or sharp deformations, and in all cases, with minimal prior knowledge and low computational cost.This work has been partially supported by the Spanish Ministry of Science and Innovation under projects RobInstruct TIN2014-58178-R and HuMoUR TIN2017-90086-R, and by a Google Faculty Award.Peer ReviewedElsevierMinisterio de Economía y Competitividad (España)Ministerio de Ciencia e Innovación (España)Consejo Superior de Investigaciones Científicas [https://ror.org/02gfc7t72]2018201820182018info:eu-repo/semantics/articlehttp://purl.org/coar/resource_type/c_6501Postprintinfo:eu-repo/semantics/acceptedVersionhttp://hdl.handle.net/10261/167144reponame:DIGITAL.CSIC. Repositorio Institucional del CSICinstname:Consejo Superior de Investigaciones Científicas (CSIC)Inglés#PLACEHOLDER_PARENT_METADATA_VALUE##PLACEHOLDER_PARENT_METADATA_VALUE##PLACEHOLDER_PARENT_METADATA_VALUE#info:eu-repo/grantAgreement/MINECO/Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016/TIN2014-58178-Rinfo:eu-repo/grantAgreement/MINECO/Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016/TIN2017-90086-Rinfo:eu-repo/grantAgreement/MINECO/Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016/TIN2017-90086-Rhttps://doi.org/10.1016/j.cviu.2018.01.002Síinfo:eu-repo/semantics/openAccessoai:digital.csic.es:10261/1671442026-05-22T06:33:51Z
dc.title.none.fl_str_mv A scalable, efficient, and accurate solution to non-rigid structure from motion
title A scalable, efficient, and accurate solution to non-rigid structure from motion
spellingShingle A scalable, efficient, and accurate solution to non-rigid structure from motion
Agudo Martínez, Antonio
Factorization
Non-rigid structure from motion
Low-rank representation
Time-varying scenes
Probabilistic trajectory space
title_short A scalable, efficient, and accurate solution to non-rigid structure from motion
title_full A scalable, efficient, and accurate solution to non-rigid structure from motion
title_fullStr A scalable, efficient, and accurate solution to non-rigid structure from motion
title_full_unstemmed A scalable, efficient, and accurate solution to non-rigid structure from motion
title_sort A scalable, efficient, and accurate solution to non-rigid structure from motion
dc.creator.none.fl_str_mv Agudo Martínez, Antonio
Moreno-Noguer, Francesc
author Agudo Martínez, Antonio
author_facet Agudo Martínez, Antonio
Moreno-Noguer, Francesc
author_role author
author2 Moreno-Noguer, Francesc
author2_role author
dc.contributor.none.fl_str_mv Ministerio de Economía y Competitividad (España)
Ministerio de Ciencia e Innovación (España)
Consejo Superior de Investigaciones Científicas [https://ror.org/02gfc7t72]
dc.subject.none.fl_str_mv Factorization
Non-rigid structure from motion
Low-rank representation
Time-varying scenes
Probabilistic trajectory space
topic Factorization
Non-rigid structure from motion
Low-rank representation
Time-varying scenes
Probabilistic trajectory space
description Most Non-Rigid Structure from Motion (NRSfM) solutions are based on factorization approaches that allow reconstructing objects parameterized by a sparse set of 3D points. These solutions, however, are low resolution and generally, they do not scale well to more than a few tens of points. While there have been recent attempts at bringing NRSfM to a dense domain, using for instance variational formulations, these are computationally demanding alternatives which require certain spatial continuity of the data, preventing their use for articulated shapes with large deformations or situations with multiple discontinuous objects. In this paper, we propose incorporating existing point trajectory low-rank models into a probabilistic framework for matrix normal distributions. With this formalism, we can then simultaneously learn shape and pose parameters using expectation maximization, and easily exploit additional priors such as known point correlations. While similar frameworks have been used before to model distributions over shapes, here we show that formulating the problem in terms of distributions over trajectories brings remarkable improvements, especially in generality and efficiency. We evaluate the proposed approach in a variety of scenarios including one or multiple objects, sparse or dense reconstructions, missing observations, mild or sharp deformations, and in all cases, with minimal prior knowledge and low computational cost.
publishDate 2018
dc.date.none.fl_str_mv 2018
2018
2018
2018
dc.type.none.fl_str_mv info:eu-repo/semantics/article
http://purl.org/coar/resource_type/c_6501
Postprint
info:eu-repo/semantics/acceptedVersion
format article
status_str acceptedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/10261/167144
url http://hdl.handle.net/10261/167144
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv #PLACEHOLDER_PARENT_METADATA_VALUE#
#PLACEHOLDER_PARENT_METADATA_VALUE#
#PLACEHOLDER_PARENT_METADATA_VALUE#
info:eu-repo/grantAgreement/MINECO/Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016/TIN2014-58178-R
info:eu-repo/grantAgreement/MINECO/Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016/TIN2017-90086-R
info:eu-repo/grantAgreement/MINECO/Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016/TIN2017-90086-R
https://doi.org/10.1016/j.cviu.2018.01.002
Sí
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:DIGITAL.CSIC. Repositorio Institucional del CSIC
instname:Consejo Superior de Investigaciones Científicas (CSIC)
instname_str Consejo Superior de Investigaciones Científicas (CSIC)
reponame_str DIGITAL.CSIC. Repositorio Institucional del CSIC
collection DIGITAL.CSIC. Repositorio Institucional del CSIC
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869414688367837184
score 15,198674

Similares en LA Referencia