A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes

Diego Thomas, Akihiro Sugimoto

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

1 Citation (Scopus)

Abstract

The frame-to-global-model approach is widely used for accurate 3D modeling from sequences of RGB-D images. Because still no perfect camera tracking system exists, the accumulation of small errors generated when registering and integrating successive RGB-D images causes deformations of the 3D model being built up. In particular, the deformations become significant when the scale of the scene to model is large. To tackle this problem, we propose a two-stage strategy to build in details a large-scale 3D model with minimal deformations where the first stage creates accurate small-scale 3D scenes in real-time from short subsequences of RGB-D images while the second stage re-organises all the results from the first stage in a geometrically consistent manner to reduce deformations as much as possible. By employing planar patches as the 3D scene representation, our proposed method runs in real-time to build accurate 3D models with minimal deformations even for large-scale scenes. Our experiments using real data confirm the effectiveness of our proposed method.

Original languageEnglish
Title of host publicationComputer Vision - ECCV 2014 Workshops, Proceedings
EditorsMichael M. Bronstein, Carsten Rother, Lourdes Agapito
PublisherSpringer Verlag
Pages428-442
Number of pages15
ISBN (Electronic)9783319161778
DOIs
Publication statusPublished - Jan 1 2015
Externally publishedYes
Event13th European Conference on Computer Vision, ECCV 2014 - Zurich, Switzerland
Duration: Sep 6 2014Sep 12 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8925
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other13th European Conference on Computer Vision, ECCV 2014
CountrySwitzerland
CityZurich
Period9/6/149/12/14

Fingerprint

3D Reconstruction
Real-time
3D Model
3D Modeling
Tracking System
Subsequence
Patch
Camera
Cameras
Strategy
Model
Experiment
Experiments

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Thomas, D., & Sugimoto, A. (2015). A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes. In M. M. Bronstein, C. Rother, & L. Agapito (Eds.), Computer Vision - ECCV 2014 Workshops, Proceedings (pp. 428-442). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8925). Springer Verlag. https://doi.org/10.1007/978-3-319-16178-5_30

A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes. / Thomas, Diego; Sugimoto, Akihiro.

Computer Vision - ECCV 2014 Workshops, Proceedings. ed. / Michael M. Bronstein; Carsten Rother; Lourdes Agapito. Springer Verlag, 2015. p. 428-442 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8925).

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

Thomas, D & Sugimoto, A 2015, A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes. in MM Bronstein, C Rother & L Agapito (eds), Computer Vision - ECCV 2014 Workshops, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8925, Springer Verlag, pp. 428-442, 13th European Conference on Computer Vision, ECCV 2014, Zurich, Switzerland, 9/6/14. https://doi.org/10.1007/978-3-319-16178-5_30
Thomas D, Sugimoto A. A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes. In Bronstein MM, Rother C, Agapito L, editors, Computer Vision - ECCV 2014 Workshops, Proceedings. Springer Verlag. 2015. p. 428-442. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16178-5_30
Thomas, Diego ; Sugimoto, Akihiro. / A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes. Computer Vision - ECCV 2014 Workshops, Proceedings. editor / Michael M. Bronstein ; Carsten Rother ; Lourdes Agapito. Springer Verlag, 2015. pp. 428-442 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{51efbf73d5604f8bb626ccb1e28eb357,
title = "A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes",
abstract = "The frame-to-global-model approach is widely used for accurate 3D modeling from sequences of RGB-D images. Because still no perfect camera tracking system exists, the accumulation of small errors generated when registering and integrating successive RGB-D images causes deformations of the 3D model being built up. In particular, the deformations become significant when the scale of the scene to model is large. To tackle this problem, we propose a two-stage strategy to build in details a large-scale 3D model with minimal deformations where the first stage creates accurate small-scale 3D scenes in real-time from short subsequences of RGB-D images while the second stage re-organises all the results from the first stage in a geometrically consistent manner to reduce deformations as much as possible. By employing planar patches as the 3D scene representation, our proposed method runs in real-time to build accurate 3D models with minimal deformations even for large-scale scenes. Our experiments using real data confirm the effectiveness of our proposed method.",
author = "Diego Thomas and Akihiro Sugimoto",
year = "2015",
month = "1",
day = "1",
doi = "10.1007/978-3-319-16178-5_30",
language = "English",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "428--442",
editor = "Bronstein, {Michael M.} and Carsten Rother and Lourdes Agapito",
booktitle = "Computer Vision - ECCV 2014 Workshops, Proceedings",
address = "Germany",

}

TY - GEN

T1 - A two-stage strategy for real-time dense 3D reconstruction of large-scale scenes

AU - Thomas, Diego

AU - Sugimoto, Akihiro

PY - 2015/1/1

Y1 - 2015/1/1

N2 - The frame-to-global-model approach is widely used for accurate 3D modeling from sequences of RGB-D images. Because still no perfect camera tracking system exists, the accumulation of small errors generated when registering and integrating successive RGB-D images causes deformations of the 3D model being built up. In particular, the deformations become significant when the scale of the scene to model is large. To tackle this problem, we propose a two-stage strategy to build in details a large-scale 3D model with minimal deformations where the first stage creates accurate small-scale 3D scenes in real-time from short subsequences of RGB-D images while the second stage re-organises all the results from the first stage in a geometrically consistent manner to reduce deformations as much as possible. By employing planar patches as the 3D scene representation, our proposed method runs in real-time to build accurate 3D models with minimal deformations even for large-scale scenes. Our experiments using real data confirm the effectiveness of our proposed method.

AB - The frame-to-global-model approach is widely used for accurate 3D modeling from sequences of RGB-D images. Because still no perfect camera tracking system exists, the accumulation of small errors generated when registering and integrating successive RGB-D images causes deformations of the 3D model being built up. In particular, the deformations become significant when the scale of the scene to model is large. To tackle this problem, we propose a two-stage strategy to build in details a large-scale 3D model with minimal deformations where the first stage creates accurate small-scale 3D scenes in real-time from short subsequences of RGB-D images while the second stage re-organises all the results from the first stage in a geometrically consistent manner to reduce deformations as much as possible. By employing planar patches as the 3D scene representation, our proposed method runs in real-time to build accurate 3D models with minimal deformations even for large-scale scenes. Our experiments using real data confirm the effectiveness of our proposed method.

UR - http://www.scopus.com/inward/record.url?scp=84925303157&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84925303157&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-16178-5_30

DO - 10.1007/978-3-319-16178-5_30

M3 - Conference contribution

AN - SCOPUS:84925303157

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

SP - 428

EP - 442

BT - Computer Vision - ECCV 2014 Workshops, Proceedings

A2 - Bronstein, Michael M.

A2 - Rother, Carsten

A2 - Agapito, Lourdes

PB - Springer Verlag

ER -