Motion estimation for regions of reflections through layer separation

Mohamed Elgharib, Francois Pitite, Anil Kokaram

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Regions of reflections contain two semi-transparent layers moving over each other. This generates two motion vectors per pel. Current multiple motion estimators either extend the usual brightness consistency assumption to two motions or are based on the Fourier phase shift relationship. Both approaches assume constant motion over at least three frames. As a result they can not handle temporally active motion due to camera shake or acceleration. This paper proposes a new approach for multiple motion estimation by modeling the correct motions as the ones generating the best layer separation of the examined reflection. A Bayesian framework is proposed which then admits a solution using candidate motions generated from KLT trajectories and a layer separation technique. We use novel temporal priors and our results show handling of strong motion inconsistencies and improvements over previous work.

Original languageEnglish
Title of host publicationProceedings - 2011 Conference for Visual Media Production, CVMP 2011
Pages49-58
Number of pages10
DOIs
Publication statusPublished - 1 Dec 2011
Externally publishedYes
Event8th European Conference on Visual Media Production, CVMP 2011 - London, United Kingdom
Duration: 16 Nov 201117 Nov 2011

Other

Other8th European Conference on Visual Media Production, CVMP 2011
CountryUnited Kingdom
CityLondon
Period16/11/1117/11/11

Fingerprint

Layer
Modeling
Brightness
Inconsistency
Trajectory

Keywords

  • Bayesian
  • Intrinsic
  • KLT
  • Multiple Motions
  • Separation

ASJC Scopus subject areas

  • Visual Arts and Performing Arts

Cite this

Elgharib, M., Pitite, F., & Kokaram, A. (2011). Motion estimation for regions of reflections through layer separation. In Proceedings - 2011 Conference for Visual Media Production, CVMP 2011 (pp. 49-58). [6103275] https://doi.org/10.1109/CVMP.2011.12

Motion estimation for regions of reflections through layer separation. / Elgharib, Mohamed; Pitite, Francois; Kokaram, Anil.

Proceedings - 2011 Conference for Visual Media Production, CVMP 2011. 2011. p. 49-58 6103275.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Elgharib, M, Pitite, F & Kokaram, A 2011, Motion estimation for regions of reflections through layer separation. in Proceedings - 2011 Conference for Visual Media Production, CVMP 2011., 6103275, pp. 49-58, 8th European Conference on Visual Media Production, CVMP 2011, London, United Kingdom, 16/11/11. https://doi.org/10.1109/CVMP.2011.12
Elgharib M, Pitite F, Kokaram A. Motion estimation for regions of reflections through layer separation. In Proceedings - 2011 Conference for Visual Media Production, CVMP 2011. 2011. p. 49-58. 6103275 https://doi.org/10.1109/CVMP.2011.12
Elgharib, Mohamed ; Pitite, Francois ; Kokaram, Anil. / Motion estimation for regions of reflections through layer separation. Proceedings - 2011 Conference for Visual Media Production, CVMP 2011. 2011. pp. 49-58
@inproceedings{6571c83b3f51427687928163d38414ca,
title = "Motion estimation for regions of reflections through layer separation",
abstract = "Regions of reflections contain two semi-transparent layers moving over each other. This generates two motion vectors per pel. Current multiple motion estimators either extend the usual brightness consistency assumption to two motions or are based on the Fourier phase shift relationship. Both approaches assume constant motion over at least three frames. As a result they can not handle temporally active motion due to camera shake or acceleration. This paper proposes a new approach for multiple motion estimation by modeling the correct motions as the ones generating the best layer separation of the examined reflection. A Bayesian framework is proposed which then admits a solution using candidate motions generated from KLT trajectories and a layer separation technique. We use novel temporal priors and our results show handling of strong motion inconsistencies and improvements over previous work.",
keywords = "Bayesian, Intrinsic, KLT, Multiple Motions, Separation",
author = "Mohamed Elgharib and Francois Pitite and Anil Kokaram",
year = "2011",
month = "12",
day = "1",
doi = "10.1109/CVMP.2011.12",
language = "English",
isbn = "9780769546216",
pages = "49--58",
booktitle = "Proceedings - 2011 Conference for Visual Media Production, CVMP 2011",

}

TY - GEN

T1 - Motion estimation for regions of reflections through layer separation

AU - Elgharib, Mohamed

AU - Pitite, Francois

AU - Kokaram, Anil

PY - 2011/12/1

Y1 - 2011/12/1

N2 - Regions of reflections contain two semi-transparent layers moving over each other. This generates two motion vectors per pel. Current multiple motion estimators either extend the usual brightness consistency assumption to two motions or are based on the Fourier phase shift relationship. Both approaches assume constant motion over at least three frames. As a result they can not handle temporally active motion due to camera shake or acceleration. This paper proposes a new approach for multiple motion estimation by modeling the correct motions as the ones generating the best layer separation of the examined reflection. A Bayesian framework is proposed which then admits a solution using candidate motions generated from KLT trajectories and a layer separation technique. We use novel temporal priors and our results show handling of strong motion inconsistencies and improvements over previous work.

AB - Regions of reflections contain two semi-transparent layers moving over each other. This generates two motion vectors per pel. Current multiple motion estimators either extend the usual brightness consistency assumption to two motions or are based on the Fourier phase shift relationship. Both approaches assume constant motion over at least three frames. As a result they can not handle temporally active motion due to camera shake or acceleration. This paper proposes a new approach for multiple motion estimation by modeling the correct motions as the ones generating the best layer separation of the examined reflection. A Bayesian framework is proposed which then admits a solution using candidate motions generated from KLT trajectories and a layer separation technique. We use novel temporal priors and our results show handling of strong motion inconsistencies and improvements over previous work.

KW - Bayesian

KW - Intrinsic

KW - KLT

KW - Multiple Motions

KW - Separation

UR - http://www.scopus.com/inward/record.url?scp=84855857836&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84855857836&partnerID=8YFLogxK

U2 - 10.1109/CVMP.2011.12

DO - 10.1109/CVMP.2011.12

M3 - Conference contribution

SN - 9780769546216

SP - 49

EP - 58

BT - Proceedings - 2011 Conference for Visual Media Production, CVMP 2011

ER -