Variational multi-valued velocity field estimation for transparent sequences.
Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt
Standard
Variational multi-valued velocity field estimation for transparent sequences. / Ramírez-Manzanares, Alonso; Rivera, Mariano; Kornprobst, Pierre; Lauze, Francois Bernard.
I: Journal of Mathematical Imaging and Vision, Bind 40, Nr. 3, 2011, s. 285-304.Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - JOUR
T1 - Variational multi-valued velocity field estimation for transparent sequences.
AU - Ramírez-Manzanares, Alonso
AU - Rivera, Mariano
AU - Kornprobst, Pierre
AU - Lauze, Francois Bernard
PY - 2011
Y1 - 2011
N2 - Motion estimation in sequences with transparenciesis an important problem in robotics and medicalimaging applications. In this work we propose avariational approach for estimating multi-valued velocityfields in transparent sequences. Starting from existinglocal motion estimators, we derive a variationalmodel for integrating in space and time such a localinformation in order to obtain a robust estimation ofthe multi-valued velocity field. With this approach, wecan indeed estimate multi-valued velocity fields whichare not necessarily piecewise constant on a layer –eachlayer can evolve according to a non-parametric opticalflow. We show how our approach outperforms existingmethods; and we illustrate its capabilities on challengingexperiments on both synthetic and real sequences.
AB - Motion estimation in sequences with transparenciesis an important problem in robotics and medicalimaging applications. In this work we propose avariational approach for estimating multi-valued velocityfields in transparent sequences. Starting from existinglocal motion estimators, we derive a variationalmodel for integrating in space and time such a localinformation in order to obtain a robust estimation ofthe multi-valued velocity field. With this approach, wecan indeed estimate multi-valued velocity fields whichare not necessarily piecewise constant on a layer –eachlayer can evolve according to a non-parametric opticalflow. We show how our approach outperforms existingmethods; and we illustrate its capabilities on challengingexperiments on both synthetic and real sequences.
KW - Faculty of Science
KW - transparent optical flow
KW - Image regularization
KW - Multiple motions
KW - RDK sequences
U2 - 10.1007/s10851-011-0260-8
DO - 10.1007/s10851-011-0260-8
M3 - Journal article
VL - 40
SP - 285
EP - 304
JO - Journal of Mathematical Imaging and Vision
JF - Journal of Mathematical Imaging and Vision
SN - 0924-9907
IS - 3
ER -
ID: 32338422