Incremental perspective motion model for rigid and non-rigid motion separation

Tzung Heng Lai, Te Hsun Wang, James Jenn-Jier Lien

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

Abstract

Motion extraction is an essential work in facial expression analysis because facial expression usually experiences rigid head rotation and non-rigid facial expression simultaneously. We developed a system to separate non-rigid motion from large rigid motion over an image sequence based on the incremental perspective motion model. Since the parameters of this motion model are able to not only represnt the global rigid motion but also localize die non-rigid motion, thus this motion model overcomes the limitations of existing memods, the affine model and the 8-parameter perspective projection model, in large head rotation angles. In addition, since the gradient descent approach is susceptible to local minimum during the motion parameter estimation process, a multi-resolution approach is applied to optimize initial values of parameters at the coarse level. Finally, the experimental result shows that our model has promising performance of separating non-rigid motion from rigid motion.

Original languageEnglish
Title of host publicationAdvances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings
Pages613-624
Number of pages12
Publication statusPublished - 2007 Dec 1
Event2nd IEEE Pacific Rim Symposium on Video and Image Technology, PSIVT 2007 - Santiago, Chile
Duration: 2007 Dec 172007 Dec 19

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4872 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other2nd IEEE Pacific Rim Symposium on Video and Image Technology, PSIVT 2007
CountryChile
CitySantiago
Period07-12-1707-12-19

Fingerprint

Motion
Facial Expression
Model
Parameter estimation
Gradient Descent
Motion Estimation
Image Sequence
Multiresolution
Local Minima
Parameter Estimation
Die
Optimise
Projection
Angle
Experimental Results

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Lai, T. H., Wang, T. H., & Lien, J. J-J. (2007). Incremental perspective motion model for rigid and non-rigid motion separation. In Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings (pp. 613-624). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4872 LNCS).
Lai, Tzung Heng ; Wang, Te Hsun ; Lien, James Jenn-Jier. / Incremental perspective motion model for rigid and non-rigid motion separation. Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings. 2007. pp. 613-624 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{a103fab33394451daf7be3529dcc5b00,
title = "Incremental perspective motion model for rigid and non-rigid motion separation",
abstract = "Motion extraction is an essential work in facial expression analysis because facial expression usually experiences rigid head rotation and non-rigid facial expression simultaneously. We developed a system to separate non-rigid motion from large rigid motion over an image sequence based on the incremental perspective motion model. Since the parameters of this motion model are able to not only represnt the global rigid motion but also localize die non-rigid motion, thus this motion model overcomes the limitations of existing memods, the affine model and the 8-parameter perspective projection model, in large head rotation angles. In addition, since the gradient descent approach is susceptible to local minimum during the motion parameter estimation process, a multi-resolution approach is applied to optimize initial values of parameters at the coarse level. Finally, the experimental result shows that our model has promising performance of separating non-rigid motion from rigid motion.",
author = "Lai, {Tzung Heng} and Wang, {Te Hsun} and Lien, {James Jenn-Jier}",
year = "2007",
month = "12",
day = "1",
language = "English",
isbn = "9783540771289",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "613--624",
booktitle = "Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings",

}

Lai, TH, Wang, TH & Lien, JJ-J 2007, Incremental perspective motion model for rigid and non-rigid motion separation. in Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4872 LNCS, pp. 613-624, 2nd IEEE Pacific Rim Symposium on Video and Image Technology, PSIVT 2007, Santiago, Chile, 07-12-17.

Incremental perspective motion model for rigid and non-rigid motion separation. / Lai, Tzung Heng; Wang, Te Hsun; Lien, James Jenn-Jier.

Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings. 2007. p. 613-624 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4872 LNCS).

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

TY - GEN

T1 - Incremental perspective motion model for rigid and non-rigid motion separation

AU - Lai, Tzung Heng

AU - Wang, Te Hsun

AU - Lien, James Jenn-Jier

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Motion extraction is an essential work in facial expression analysis because facial expression usually experiences rigid head rotation and non-rigid facial expression simultaneously. We developed a system to separate non-rigid motion from large rigid motion over an image sequence based on the incremental perspective motion model. Since the parameters of this motion model are able to not only represnt the global rigid motion but also localize die non-rigid motion, thus this motion model overcomes the limitations of existing memods, the affine model and the 8-parameter perspective projection model, in large head rotation angles. In addition, since the gradient descent approach is susceptible to local minimum during the motion parameter estimation process, a multi-resolution approach is applied to optimize initial values of parameters at the coarse level. Finally, the experimental result shows that our model has promising performance of separating non-rigid motion from rigid motion.

AB - Motion extraction is an essential work in facial expression analysis because facial expression usually experiences rigid head rotation and non-rigid facial expression simultaneously. We developed a system to separate non-rigid motion from large rigid motion over an image sequence based on the incremental perspective motion model. Since the parameters of this motion model are able to not only represnt the global rigid motion but also localize die non-rigid motion, thus this motion model overcomes the limitations of existing memods, the affine model and the 8-parameter perspective projection model, in large head rotation angles. In addition, since the gradient descent approach is susceptible to local minimum during the motion parameter estimation process, a multi-resolution approach is applied to optimize initial values of parameters at the coarse level. Finally, the experimental result shows that our model has promising performance of separating non-rigid motion from rigid motion.

UR - http://www.scopus.com/inward/record.url?scp=38349063291&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38349063291&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540771289

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 613

EP - 624

BT - Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings

ER -

Lai TH, Wang TH, Lien JJ-J. Incremental perspective motion model for rigid and non-rigid motion separation. In Advances in Image and Video Technology - Second Pacific Rim Symposium, PSIVT 2007, Proceedings. 2007. p. 613-624. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).