Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments

Clement Godard, Peter Hedman, Wenbin Li, Gabriel J. Brostow

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

15 Citations (Scopus)

Abstract

Reconstructing the surface of highly specular objects is a challenging task. The shapes of diffuse and rough specular objects can be captured in an uncontrolled setting using consumer equipment. In contrast, highly specular objects have previously deterred capture in uncontrolled environments and have only been reconstructed using tailor-made hardware. We propose a method to reconstruct such objects in uncontrolled environments using only commodity hardware. As input, our method expects multi-view photographs of the specular object, its silhouettes and an environment map of its surroundings. We compare the reflected colors in the photographs with the ones in the environment to form probability distributions over the surface normals. As the effect of inter-reflections cannot be ignored for highly specular objects, we explicitly model them when forming the probability distributions. We recover the shape of the object in an iterative process where we alternate between estimating normals and updating the shape of the object to better explain these normals. We run experiments on both synthetic and real-world data, that show our method is robust and produces accurate reconstructions with as few as 25 input photographs.

Original languageEnglish
Title of host publicationProceedings - 2015 International Conference on 3D Vision, 3DV 2015
EditorsMichael Brown, Jana Kosecka, Christian Theobalt
PublisherIEEE
Pages19-27
Number of pages9
ISBN (Electronic)9781467383325
DOIs
Publication statusPublished - 20 Nov 2015
Event2015 International Conference on 3D Vision, 3DV 2015 - Lyon, France
Duration: 19 Oct 201522 Oct 2015

Conference

Conference2015 International Conference on 3D Vision, 3DV 2015
CountryFrance
CityLyon
Period19/10/1522/10/15

Keywords

  • multi-view reconstruction
  • specular surface
  • surface reconstruction

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Godard, C., Hedman, P., Li, W., & Brostow, G. J. (2015). Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments. In M. Brown, J. Kosecka, & C. Theobalt (Eds.), Proceedings - 2015 International Conference on 3D Vision, 3DV 2015 (pp. 19-27). [7335465] IEEE. https://doi.org/10.1109/3DV.2015.10

Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments. / Godard, Clement; Hedman, Peter; Li, Wenbin; Brostow, Gabriel J.

Proceedings - 2015 International Conference on 3D Vision, 3DV 2015. ed. / Michael Brown; Jana Kosecka; Christian Theobalt. IEEE, 2015. p. 19-27 7335465.

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

Godard, C, Hedman, P, Li, W & Brostow, GJ 2015, Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments. in M Brown, J Kosecka & C Theobalt (eds), Proceedings - 2015 International Conference on 3D Vision, 3DV 2015., 7335465, IEEE, pp. 19-27, 2015 International Conference on 3D Vision, 3DV 2015, Lyon, France, 19/10/15. https://doi.org/10.1109/3DV.2015.10
Godard C, Hedman P, Li W, Brostow GJ. Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments. In Brown M, Kosecka J, Theobalt C, editors, Proceedings - 2015 International Conference on 3D Vision, 3DV 2015. IEEE. 2015. p. 19-27. 7335465 https://doi.org/10.1109/3DV.2015.10
Godard, Clement ; Hedman, Peter ; Li, Wenbin ; Brostow, Gabriel J. / Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments. Proceedings - 2015 International Conference on 3D Vision, 3DV 2015. editor / Michael Brown ; Jana Kosecka ; Christian Theobalt. IEEE, 2015. pp. 19-27
@inproceedings{732498043b1347b0ad968b37b10e30f3,
title = "Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments",
abstract = "Reconstructing the surface of highly specular objects is a challenging task. The shapes of diffuse and rough specular objects can be captured in an uncontrolled setting using consumer equipment. In contrast, highly specular objects have previously deterred capture in uncontrolled environments and have only been reconstructed using tailor-made hardware. We propose a method to reconstruct such objects in uncontrolled environments using only commodity hardware. As input, our method expects multi-view photographs of the specular object, its silhouettes and an environment map of its surroundings. We compare the reflected colors in the photographs with the ones in the environment to form probability distributions over the surface normals. As the effect of inter-reflections cannot be ignored for highly specular objects, we explicitly model them when forming the probability distributions. We recover the shape of the object in an iterative process where we alternate between estimating normals and updating the shape of the object to better explain these normals. We run experiments on both synthetic and real-world data, that show our method is robust and produces accurate reconstructions with as few as 25 input photographs.",
keywords = "multi-view reconstruction, specular surface, surface reconstruction",
author = "Clement Godard and Peter Hedman and Wenbin Li and Brostow, {Gabriel J.}",
year = "2015",
month = "11",
day = "20",
doi = "10.1109/3DV.2015.10",
language = "English",
pages = "19--27",
editor = "Michael Brown and Jana Kosecka and Christian Theobalt",
booktitle = "Proceedings - 2015 International Conference on 3D Vision, 3DV 2015",
publisher = "IEEE",
address = "USA United States",

}

TY - GEN

T1 - Multi-view Reconstruction of Highly Specular Surfaces in Uncontrolled Environments

AU - Godard, Clement

AU - Hedman, Peter

AU - Li, Wenbin

AU - Brostow, Gabriel J.

PY - 2015/11/20

Y1 - 2015/11/20

N2 - Reconstructing the surface of highly specular objects is a challenging task. The shapes of diffuse and rough specular objects can be captured in an uncontrolled setting using consumer equipment. In contrast, highly specular objects have previously deterred capture in uncontrolled environments and have only been reconstructed using tailor-made hardware. We propose a method to reconstruct such objects in uncontrolled environments using only commodity hardware. As input, our method expects multi-view photographs of the specular object, its silhouettes and an environment map of its surroundings. We compare the reflected colors in the photographs with the ones in the environment to form probability distributions over the surface normals. As the effect of inter-reflections cannot be ignored for highly specular objects, we explicitly model them when forming the probability distributions. We recover the shape of the object in an iterative process where we alternate between estimating normals and updating the shape of the object to better explain these normals. We run experiments on both synthetic and real-world data, that show our method is robust and produces accurate reconstructions with as few as 25 input photographs.

AB - Reconstructing the surface of highly specular objects is a challenging task. The shapes of diffuse and rough specular objects can be captured in an uncontrolled setting using consumer equipment. In contrast, highly specular objects have previously deterred capture in uncontrolled environments and have only been reconstructed using tailor-made hardware. We propose a method to reconstruct such objects in uncontrolled environments using only commodity hardware. As input, our method expects multi-view photographs of the specular object, its silhouettes and an environment map of its surroundings. We compare the reflected colors in the photographs with the ones in the environment to form probability distributions over the surface normals. As the effect of inter-reflections cannot be ignored for highly specular objects, we explicitly model them when forming the probability distributions. We recover the shape of the object in an iterative process where we alternate between estimating normals and updating the shape of the object to better explain these normals. We run experiments on both synthetic and real-world data, that show our method is robust and produces accurate reconstructions with as few as 25 input photographs.

KW - multi-view reconstruction

KW - specular surface

KW - surface reconstruction

UR - http://www.scopus.com/inward/record.url?scp=84961732811&partnerID=8YFLogxK

U2 - 10.1109/3DV.2015.10

DO - 10.1109/3DV.2015.10

M3 - Conference contribution

SP - 19

EP - 27

BT - Proceedings - 2015 International Conference on 3D Vision, 3DV 2015

A2 - Brown, Michael

A2 - Kosecka, Jana

A2 - Theobalt, Christian

PB - IEEE

ER -