Large scale multi-view stereopsis evaluation

Rasmus Jensen*, Anders Dahl, George Vogiatzis, Engil Tola, Henrik Aanæs

*Corresponding author for this work

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

Abstract

The seminal multiple view stereo benchmark evaluations from Middlebury and by Strecha et al. have played a major role in propelling the development of multi-view stereopsis methodology. Although seminal, these benchmark datasets are limited in scope with few reference scenes. Here, we try to take these works a step further by proposing a new multi-view stereo dataset, which is an order of magnitude larger in number of scenes and with a significant increase in diversity. Specifically, we propose a dataset containing 80 scenes of large variability. Each scene consists of 49 or 64 accurate camera positions and reference structured light scans, all acquired by a 6-axis industrial robot. To apply this dataset we propose an extension of the evaluation protocol from the Middlebury evaluation, reflecting the more complex geometry of some of our scenes. The proposed dataset is used to evaluate the state of the art multiview stereo algorithms of Tola et al., Campbell et al. and Furukawa et al. Hereby we demonstrate the usability of the dataset as well as gain insight into the workings and challenges of multi-view stereopsis. Through these experiments we empirically validate some of the central hypotheses of multi-view stereopsis, as well as determining and reaffirming some of the central challenges.

Original languageEnglish
Title of host publicationProceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014
Place of PublicationPiscataway, NJ (US)
PublisherIEEE
Pages406-413
Number of pages8
ISBN (Print)978-1-4799-5117-8
DOIs
Publication statusPublished - 31 Dec 2014
Event27th IEEE Conference on Computer Vision and Pattern Recognition - Columbus, OH, United States
Duration: 23 Jun 201428 Jun 2014

Publication series

NameIEEE conference publications
PublisherIEEE
ISSN (Print)1063-6919

Conference

Conference27th IEEE Conference on Computer Vision and Pattern Recognition
Abbreviated titleCVPR 2014
CountryUnited States
CityColumbus, OH
Period23/06/1428/06/14

Fingerprint

Industrial robots
Cameras
Geometry
Experiments

Keywords

  • multi view stereopsis
  • structured light
  • surface reconstruction

Cite this

Jensen, R., Dahl, A., Vogiatzis, G., Tola, E., & Aanæs, H. (2014). Large scale multi-view stereopsis evaluation. In Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014 (pp. 406-413). (IEEE conference publications). Piscataway, NJ (US): IEEE. https://doi.org/10.1109/CVPR.2014.59
Jensen, Rasmus ; Dahl, Anders ; Vogiatzis, George ; Tola, Engil ; Aanæs, Henrik. / Large scale multi-view stereopsis evaluation. Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014. Piscataway, NJ (US) : IEEE, 2014. pp. 406-413 (IEEE conference publications).
@inproceedings{ca1a437732fa4ea0b849d3993b16f3e6,
title = "Large scale multi-view stereopsis evaluation",
abstract = "The seminal multiple view stereo benchmark evaluations from Middlebury and by Strecha et al. have played a major role in propelling the development of multi-view stereopsis methodology. Although seminal, these benchmark datasets are limited in scope with few reference scenes. Here, we try to take these works a step further by proposing a new multi-view stereo dataset, which is an order of magnitude larger in number of scenes and with a significant increase in diversity. Specifically, we propose a dataset containing 80 scenes of large variability. Each scene consists of 49 or 64 accurate camera positions and reference structured light scans, all acquired by a 6-axis industrial robot. To apply this dataset we propose an extension of the evaluation protocol from the Middlebury evaluation, reflecting the more complex geometry of some of our scenes. The proposed dataset is used to evaluate the state of the art multiview stereo algorithms of Tola et al., Campbell et al. and Furukawa et al. Hereby we demonstrate the usability of the dataset as well as gain insight into the workings and challenges of multi-view stereopsis. Through these experiments we empirically validate some of the central hypotheses of multi-view stereopsis, as well as determining and reaffirming some of the central challenges.",
keywords = "multi view stereopsis, structured light, surface reconstruction",
author = "Rasmus Jensen and Anders Dahl and George Vogiatzis and Engil Tola and Henrik Aan{\ae}s",
year = "2014",
month = "12",
day = "31",
doi = "10.1109/CVPR.2014.59",
language = "English",
isbn = "978-1-4799-5117-8",
series = "IEEE conference publications",
publisher = "IEEE",
pages = "406--413",
booktitle = "Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014",
address = "United States",

}

Jensen, R, Dahl, A, Vogiatzis, G, Tola, E & Aanæs, H 2014, Large scale multi-view stereopsis evaluation. in Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014. IEEE conference publications, IEEE, Piscataway, NJ (US), pp. 406-413, 27th IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, United States, 23/06/14. https://doi.org/10.1109/CVPR.2014.59

Large scale multi-view stereopsis evaluation. / Jensen, Rasmus; Dahl, Anders; Vogiatzis, George; Tola, Engil; Aanæs, Henrik.

Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014. Piscataway, NJ (US) : IEEE, 2014. p. 406-413 (IEEE conference publications).

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

TY - GEN

T1 - Large scale multi-view stereopsis evaluation

AU - Jensen, Rasmus

AU - Dahl, Anders

AU - Vogiatzis, George

AU - Tola, Engil

AU - Aanæs, Henrik

PY - 2014/12/31

Y1 - 2014/12/31

N2 - The seminal multiple view stereo benchmark evaluations from Middlebury and by Strecha et al. have played a major role in propelling the development of multi-view stereopsis methodology. Although seminal, these benchmark datasets are limited in scope with few reference scenes. Here, we try to take these works a step further by proposing a new multi-view stereo dataset, which is an order of magnitude larger in number of scenes and with a significant increase in diversity. Specifically, we propose a dataset containing 80 scenes of large variability. Each scene consists of 49 or 64 accurate camera positions and reference structured light scans, all acquired by a 6-axis industrial robot. To apply this dataset we propose an extension of the evaluation protocol from the Middlebury evaluation, reflecting the more complex geometry of some of our scenes. The proposed dataset is used to evaluate the state of the art multiview stereo algorithms of Tola et al., Campbell et al. and Furukawa et al. Hereby we demonstrate the usability of the dataset as well as gain insight into the workings and challenges of multi-view stereopsis. Through these experiments we empirically validate some of the central hypotheses of multi-view stereopsis, as well as determining and reaffirming some of the central challenges.

AB - The seminal multiple view stereo benchmark evaluations from Middlebury and by Strecha et al. have played a major role in propelling the development of multi-view stereopsis methodology. Although seminal, these benchmark datasets are limited in scope with few reference scenes. Here, we try to take these works a step further by proposing a new multi-view stereo dataset, which is an order of magnitude larger in number of scenes and with a significant increase in diversity. Specifically, we propose a dataset containing 80 scenes of large variability. Each scene consists of 49 or 64 accurate camera positions and reference structured light scans, all acquired by a 6-axis industrial robot. To apply this dataset we propose an extension of the evaluation protocol from the Middlebury evaluation, reflecting the more complex geometry of some of our scenes. The proposed dataset is used to evaluate the state of the art multiview stereo algorithms of Tola et al., Campbell et al. and Furukawa et al. Hereby we demonstrate the usability of the dataset as well as gain insight into the workings and challenges of multi-view stereopsis. Through these experiments we empirically validate some of the central hypotheses of multi-view stereopsis, as well as determining and reaffirming some of the central challenges.

KW - multi view stereopsis

KW - structured light

KW - surface reconstruction

UR - http://www.scopus.com/inward/record.url?scp=84911414859&partnerID=8YFLogxK

UR - http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6909453

U2 - 10.1109/CVPR.2014.59

DO - 10.1109/CVPR.2014.59

M3 - Conference contribution

AN - SCOPUS:84911414859

SN - 978-1-4799-5117-8

T3 - IEEE conference publications

SP - 406

EP - 413

BT - Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014

PB - IEEE

CY - Piscataway, NJ (US)

ER -

Jensen R, Dahl A, Vogiatzis G, Tola E, Aanæs H. Large scale multi-view stereopsis evaluation. In Proceedings : 2014 IEEE Computer Society conference on Computer Vision and Pattern Recognition, CVPR 2014. Piscataway, NJ (US): IEEE. 2014. p. 406-413. (IEEE conference publications). https://doi.org/10.1109/CVPR.2014.59