LNCS 3024 Monocular 3D Reconstruction of Human
Motion in Long Action Sequences 1st edition by
Gareth Loy, Martin Eriksson, Josephine Sullivan,
Stefan Carlsson ISBN 3540219811 978-3540219811
pdf download
https://ebookball.com/product/lncs-3024-monocular-3d-
reconstruction-of-human-motion-in-long-action-sequences-1st-
edition-by-gareth-loy-martin-eriksson-josephine-sullivan-stefan-
carlsson-isbn-3540219811-978-3540219811-10088/
Explore and download more ebooks or textbooks
at ebookball.com
, Get Your Digital Files Instantly: PDF, ePub, MOBI and More
Quick Digital Downloads: PDF, ePub, MOBI and Other Formats
LNCS 3024 Morphological Operations on Matrix Valued Images 1st
edition by Bernhard Burgeth, Martin Welk, Christian Feddern, Joachim
Weickert ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-morphological-operations-
on-matrix-valued-images-1st-edition-by-bernhard-burgeth-martin-
welk-christian-feddern-joachim-weickert-
isbn-3540219811-978-3540219811-13114/
LNCS 3024 Reliable Fiducial Detection in Natural Scenes 1st edition by
David Claus, Andrew Fitzgiblon ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-reliable-fiducial-
detection-in-natural-scenes-1st-edition-by-david-claus-andrew-
fitzgiblon-isbn-3540219811-978-3540219811-11374/
LNCS 3024 Tracking People with a Sparse Network of Bearing Sensors 1st
edition by Rahimi, Dunagan, Darrell ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-tracking-people-with-a-
sparse-network-of-bearing-sensors-1st-edition-by-rahimi-dunagan-
darrell-isbn-3540219811-978-3540219811-14496/
LNCS 3024 Toward Accurate Segmentation of the LV Myocardium and
Chamber for Volumes Estimation in Gated SPECT Sequences 1st edition b
Diane Lingrand, Arnand Charnooz, Pierre Malick Koulibary, Jacques
Darcourt, Johan Montagnat ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-toward-accurate-
segmentation-of-the-lv-myocardium-and-chamber-for-volumes-
estimation-in-gated-spect-sequences-1st-edition-by-diane-
lingrand-arnand-charnooz-pierre-malick-koulibary-jacques-da/
,LNCS 3024 Seamless Image Stitching in the Gradient Domain 1st edition
by Anat Levin, Assaf Zormet, Shrmuel Peleg, Yair Weiss ISBN 3540219811
978-3540219811
https://ebookball.com/product/lncs-3024-seamless-image-stitching-
in-the-gradient-domain-1st-edition-by-anat-levin-assaf-zormet-
shrmuel-peleg-yair-weiss-isbn-3540219811-978-3540219811-13856/
LNCS 3024 Stereovision Based Head Tracking Using Color and Ellipse
Fitting in a Particle Filter 1st edition by Bogdan Kwolek ISBN
3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-stereovision-based-head-
tracking-using-color-and-ellipse-fitting-in-a-particle-
filter-1st-edition-by-bogdan-kwolek-
isbn-3540219811-978-3540219811-12932/
LNCS 3024 Transformation Invariant Embedding for Image Analysis 1st
edition by Ali Ghodsi, Jiaytian Huang, Dale Schuurmans ISBN 3540219811
978-3540219811
https://ebookball.com/product/lncs-3024-transformation-invariant-
embedding-for-image-analysis-1st-edition-by-ali-ghodsi-jiaytian-
huang-dale-schuurmans-isbn-3540219811-978-3540219811-14006/
LNCS 3024 Multiphase Dynamic Labeling for Variational Recognition-
Driven Image Segmentation 1st edition by Daniel Cremers, Nir Sochen,
Christoph Schnorr ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-multiphase-dynamic-
labeling-for-variational-recognition-driven-image-
segmentation-1st-edition-by-daniel-cremers-nir-sochen-christoph-
schnorr-isbn-3540219811-978-3540219811-11458/
LNCS 3024 - Light Field Appearance Manifolds 1st edition by Chris
Mario Chiristoudias, Louis Philippe Morency and Trevor Darrell ISBN
3540219811 Â 978-3540219811
https://ebookball.com/product/lncs-3024-light-field-appearance-
manifolds-1st-edition-by-chris-mario-chiristoudias-louis-
philippe-morency-and-trevor-darrell-
isbn-3540219811-978-3540219811-11920/
, Monocular 3D Reconstruction of Human Motion in
Long Action Sequences
Gareth Loy, Martin Eriksson, Josephine Sullivan, and Stefan Carlsson
Computational Vision & Active Perception Laboratory (CVAP),
Department of Numerical Analysis and Computer Science,
Royal Institute of Technology (KTH), S-100 44 Stockholm, Sweden
{gareth,eriksson,sullivan,stefanc}@nada.kth.se
Abstract. A novel algorithm is presented for the 3D reconstruction of human ac-
tion in long (> 30 second) monocular image sequences. A sequence is represented
by a small set of automatically found representative keyframes. The skeletal joint
positions are manually located in each keyframe and mapped to all other frames
in the sequence. For each keyframe a 3D key pose is created, and interpolation
between these 3D body poses, together with the incorporation of limb length and
symmetry constraints, provides a smooth initial approximation of the 3D motion.
This is then fitted to the image data to generate a realistic 3D reconstruction. The
degree of manual input required is controlled by the diversity of the sequence’s
content. Sports’ footage is ideally suited to this approach as it frequently con-
tains a limited number of repeated actions. Our method is demonstrated on a long
(36 second) sequence of a woman playing tennis filmed with a non-stationary
camera. This sequence required manual initialisation on < 1.5% of the frames,
and demonstrates that the system can deal with very rapid motion, severe self-
occlusions, motion blur and clutter occurring over several concurrent frames. The
monocular 3D reconstruction is verified by synthesising a view from the per-
spective of a ’ground truth’ reference camera, and the result is seen to provide a
qualitatively accurate 3D reconstruction of the motion.
1 Introduction
This paper addresses the challenge of generating a qualitatively accurate 3D reconstruc-
tion of the actions performed by an individual in a long (∼30 second) monocular image
sequence. It is assumed the individual is not wearing any special reflective markers or
clothing. Any solution must be able to cope with the multitude of difficulties that may
arise over several concurrent frames: severe self-occlusion, unreliability of methods for
limb and joint detection, motion blur, and the inherent ambiguities in reconstructing rigid
links from monocular images [15]. Until now, the only approach guaranteed to produce
a complete and accurate reconstruction in such circumstances is: for each frame in the
sequence, manually locate the skeletal joints and perform 3D reconstruction using the
method of [15]. The latter involves solving the forward/backward binary ambiguity for
each rigid link by inspection and estimating the relative lengths of each limb. For very
short sequences this is a relatively painless procedure, but rapidly becomes impractical
for longer sequences.
T. Pajdla and J. Matas (Eds.): ECCV 2004, LNCS 3024, pp. 442–455, 2004.
c Springer-Verlag Berlin Heidelberg 2004
Motion in Long Action Sequences 1st edition by
Gareth Loy, Martin Eriksson, Josephine Sullivan,
Stefan Carlsson ISBN 3540219811 978-3540219811
pdf download
https://ebookball.com/product/lncs-3024-monocular-3d-
reconstruction-of-human-motion-in-long-action-sequences-1st-
edition-by-gareth-loy-martin-eriksson-josephine-sullivan-stefan-
carlsson-isbn-3540219811-978-3540219811-10088/
Explore and download more ebooks or textbooks
at ebookball.com
, Get Your Digital Files Instantly: PDF, ePub, MOBI and More
Quick Digital Downloads: PDF, ePub, MOBI and Other Formats
LNCS 3024 Morphological Operations on Matrix Valued Images 1st
edition by Bernhard Burgeth, Martin Welk, Christian Feddern, Joachim
Weickert ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-morphological-operations-
on-matrix-valued-images-1st-edition-by-bernhard-burgeth-martin-
welk-christian-feddern-joachim-weickert-
isbn-3540219811-978-3540219811-13114/
LNCS 3024 Reliable Fiducial Detection in Natural Scenes 1st edition by
David Claus, Andrew Fitzgiblon ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-reliable-fiducial-
detection-in-natural-scenes-1st-edition-by-david-claus-andrew-
fitzgiblon-isbn-3540219811-978-3540219811-11374/
LNCS 3024 Tracking People with a Sparse Network of Bearing Sensors 1st
edition by Rahimi, Dunagan, Darrell ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-tracking-people-with-a-
sparse-network-of-bearing-sensors-1st-edition-by-rahimi-dunagan-
darrell-isbn-3540219811-978-3540219811-14496/
LNCS 3024 Toward Accurate Segmentation of the LV Myocardium and
Chamber for Volumes Estimation in Gated SPECT Sequences 1st edition b
Diane Lingrand, Arnand Charnooz, Pierre Malick Koulibary, Jacques
Darcourt, Johan Montagnat ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-toward-accurate-
segmentation-of-the-lv-myocardium-and-chamber-for-volumes-
estimation-in-gated-spect-sequences-1st-edition-by-diane-
lingrand-arnand-charnooz-pierre-malick-koulibary-jacques-da/
,LNCS 3024 Seamless Image Stitching in the Gradient Domain 1st edition
by Anat Levin, Assaf Zormet, Shrmuel Peleg, Yair Weiss ISBN 3540219811
978-3540219811
https://ebookball.com/product/lncs-3024-seamless-image-stitching-
in-the-gradient-domain-1st-edition-by-anat-levin-assaf-zormet-
shrmuel-peleg-yair-weiss-isbn-3540219811-978-3540219811-13856/
LNCS 3024 Stereovision Based Head Tracking Using Color and Ellipse
Fitting in a Particle Filter 1st edition by Bogdan Kwolek ISBN
3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-stereovision-based-head-
tracking-using-color-and-ellipse-fitting-in-a-particle-
filter-1st-edition-by-bogdan-kwolek-
isbn-3540219811-978-3540219811-12932/
LNCS 3024 Transformation Invariant Embedding for Image Analysis 1st
edition by Ali Ghodsi, Jiaytian Huang, Dale Schuurmans ISBN 3540219811
978-3540219811
https://ebookball.com/product/lncs-3024-transformation-invariant-
embedding-for-image-analysis-1st-edition-by-ali-ghodsi-jiaytian-
huang-dale-schuurmans-isbn-3540219811-978-3540219811-14006/
LNCS 3024 Multiphase Dynamic Labeling for Variational Recognition-
Driven Image Segmentation 1st edition by Daniel Cremers, Nir Sochen,
Christoph Schnorr ISBN 3540219811 978-3540219811
https://ebookball.com/product/lncs-3024-multiphase-dynamic-
labeling-for-variational-recognition-driven-image-
segmentation-1st-edition-by-daniel-cremers-nir-sochen-christoph-
schnorr-isbn-3540219811-978-3540219811-11458/
LNCS 3024 - Light Field Appearance Manifolds 1st edition by Chris
Mario Chiristoudias, Louis Philippe Morency and Trevor Darrell ISBN
3540219811 Â 978-3540219811
https://ebookball.com/product/lncs-3024-light-field-appearance-
manifolds-1st-edition-by-chris-mario-chiristoudias-louis-
philippe-morency-and-trevor-darrell-
isbn-3540219811-978-3540219811-11920/
, Monocular 3D Reconstruction of Human Motion in
Long Action Sequences
Gareth Loy, Martin Eriksson, Josephine Sullivan, and Stefan Carlsson
Computational Vision & Active Perception Laboratory (CVAP),
Department of Numerical Analysis and Computer Science,
Royal Institute of Technology (KTH), S-100 44 Stockholm, Sweden
{gareth,eriksson,sullivan,stefanc}@nada.kth.se
Abstract. A novel algorithm is presented for the 3D reconstruction of human ac-
tion in long (> 30 second) monocular image sequences. A sequence is represented
by a small set of automatically found representative keyframes. The skeletal joint
positions are manually located in each keyframe and mapped to all other frames
in the sequence. For each keyframe a 3D key pose is created, and interpolation
between these 3D body poses, together with the incorporation of limb length and
symmetry constraints, provides a smooth initial approximation of the 3D motion.
This is then fitted to the image data to generate a realistic 3D reconstruction. The
degree of manual input required is controlled by the diversity of the sequence’s
content. Sports’ footage is ideally suited to this approach as it frequently con-
tains a limited number of repeated actions. Our method is demonstrated on a long
(36 second) sequence of a woman playing tennis filmed with a non-stationary
camera. This sequence required manual initialisation on < 1.5% of the frames,
and demonstrates that the system can deal with very rapid motion, severe self-
occlusions, motion blur and clutter occurring over several concurrent frames. The
monocular 3D reconstruction is verified by synthesising a view from the per-
spective of a ’ground truth’ reference camera, and the result is seen to provide a
qualitatively accurate 3D reconstruction of the motion.
1 Introduction
This paper addresses the challenge of generating a qualitatively accurate 3D reconstruc-
tion of the actions performed by an individual in a long (∼30 second) monocular image
sequence. It is assumed the individual is not wearing any special reflective markers or
clothing. Any solution must be able to cope with the multitude of difficulties that may
arise over several concurrent frames: severe self-occlusion, unreliability of methods for
limb and joint detection, motion blur, and the inherent ambiguities in reconstructing rigid
links from monocular images [15]. Until now, the only approach guaranteed to produce
a complete and accurate reconstruction in such circumstances is: for each frame in the
sequence, manually locate the skeletal joints and perform 3D reconstruction using the
method of [15]. The latter involves solving the forward/backward binary ambiguity for
each rigid link by inspection and estimating the relative lengths of each limb. For very
short sequences this is a relatively painless procedure, but rapidly becomes impractical
for longer sequences.
T. Pajdla and J. Matas (Eds.): ECCV 2004, LNCS 3024, pp. 442–455, 2004.
c Springer-Verlag Berlin Heidelberg 2004
