A Robust System for High-Quality Reconstruction of 3D Objects from Photographs

H. M. Nguyen, B. C. Wünsche, P. Delmas, C. Lutteroth, W. van der Mark

Research output: Chapter in Book/Report/Conference proceedingOther chapter contribution

1 Citation (Scopus)

Abstract

Image-based modeling is rapidly increasing in popularity since cameras are very affordable, widely available, and have a wide image acquisition range suitable for objects of vastly different size. In this chapter we describe a novel image-based modeling system, which produces high-quality 3D content automatically from a collection of unconstrained and uncalibrated 2D images. The system estimates camera parameters and a 3D scene geometry using Structure-from-Motion (SfM) and Bundle Adjustment techniques. The point cloud density of 3D scene components is enhanced by exploiting silhouette information of the scene. This hybrid approach dramatically improves the reconstruction of objects with few visual features. A high quality texture is created by parameterizing the reconstructed objects using a segmentation and charting approach, which also works for objects which are not homeomorphic to a sphere. The resulting parameter space contains one chart for each surface segment. A texture map is created by back projecting the best fitting input images onto each surface segment, and smoothly fusing them together over the corresponding chart by using graph-cut techniques. Our evaluation shows that our system is capable of reconstructing a wide range of objects in both indoor and outdoor environments.




Original languageEnglish
Title of host publicationIssues and Challenges in Artificial Intelligence
Subtitle of host publicationPart 1
EditorsZ. S. Hippe, J. L. Kulikowski, T. Mroczek, J. Wtorek
PublisherSpringer International Publishing
Pages3-15
Number of pages13
ISBN (Print)9783319068824
DOIs
Publication statusPublished - 2014

Publication series

NameStudies in Computational Intelligence
Volume559

Fingerprint

Textures
Cameras
Image acquisition
Geometry

Cite this

Nguyen, H. M., Wünsche, B. C., Delmas, P., Lutteroth, C., & van der Mark, W. (2014). A Robust System for High-Quality Reconstruction of 3D Objects from Photographs. In Z. S. Hippe, J. L. Kulikowski, T. Mroczek, & J. Wtorek (Eds.), Issues and Challenges in Artificial Intelligence: Part 1 (pp. 3-15). (Studies in Computational Intelligence; Vol. 559). Springer International Publishing. https://doi.org/10.1007/978-3-319-06883-1_1

A Robust System for High-Quality Reconstruction of 3D Objects from Photographs. / Nguyen, H. M.; Wünsche, B. C.; Delmas, P.; Lutteroth, C.; van der Mark, W.

Issues and Challenges in Artificial Intelligence: Part 1. ed. / Z. S. Hippe; J. L. Kulikowski; T. Mroczek; J. Wtorek. Springer International Publishing, 2014. p. 3-15 (Studies in Computational Intelligence; Vol. 559).

Research output: Chapter in Book/Report/Conference proceedingOther chapter contribution

Nguyen, HM, Wünsche, BC, Delmas, P, Lutteroth, C & van der Mark, W 2014, A Robust System for High-Quality Reconstruction of 3D Objects from Photographs. in ZS Hippe, JL Kulikowski, T Mroczek & J Wtorek (eds), Issues and Challenges in Artificial Intelligence: Part 1. Studies in Computational Intelligence, vol. 559, Springer International Publishing, pp. 3-15. https://doi.org/10.1007/978-3-319-06883-1_1
Nguyen HM, Wünsche BC, Delmas P, Lutteroth C, van der Mark W. A Robust System for High-Quality Reconstruction of 3D Objects from Photographs. In Hippe ZS, Kulikowski JL, Mroczek T, Wtorek J, editors, Issues and Challenges in Artificial Intelligence: Part 1. Springer International Publishing. 2014. p. 3-15. (Studies in Computational Intelligence). https://doi.org/10.1007/978-3-319-06883-1_1
Nguyen, H. M. ; Wünsche, B. C. ; Delmas, P. ; Lutteroth, C. ; van der Mark, W. / A Robust System for High-Quality Reconstruction of 3D Objects from Photographs. Issues and Challenges in Artificial Intelligence: Part 1. editor / Z. S. Hippe ; J. L. Kulikowski ; T. Mroczek ; J. Wtorek. Springer International Publishing, 2014. pp. 3-15 (Studies in Computational Intelligence).
@inbook{bb70b7c747584cd0b5359d7ca4c22acd,
title = "A Robust System for High-Quality Reconstruction of 3D Objects from Photographs",
abstract = "Image-based modeling is rapidly increasing in popularity since cameras are very affordable, widely available, and have a wide image acquisition range suitable for objects of vastly different size. In this chapter we describe a novel image-based modeling system, which produces high-quality 3D content automatically from a collection of unconstrained and uncalibrated 2D images. The system estimates camera parameters and a 3D scene geometry using Structure-from-Motion (SfM) and Bundle Adjustment techniques. The point cloud density of 3D scene components is enhanced by exploiting silhouette information of the scene. This hybrid approach dramatically improves the reconstruction of objects with few visual features. A high quality texture is created by parameterizing the reconstructed objects using a segmentation and charting approach, which also works for objects which are not homeomorphic to a sphere. The resulting parameter space contains one chart for each surface segment. A texture map is created by back projecting the best fitting input images onto each surface segment, and smoothly fusing them together over the corresponding chart by using graph-cut techniques. Our evaluation shows that our system is capable of reconstructing a wide range of objects in both indoor and outdoor environments.",
author = "Nguyen, {H. M.} and W{\"u}nsche, {B. C.} and P. Delmas and C. Lutteroth and {van der Mark}, W.",
year = "2014",
doi = "10.1007/978-3-319-06883-1_1",
language = "English",
isbn = "9783319068824",
series = "Studies in Computational Intelligence",
publisher = "Springer International Publishing",
pages = "3--15",
editor = "Hippe, {Z. S.} and Kulikowski, {J. L.} and T. Mroczek and J. Wtorek",
booktitle = "Issues and Challenges in Artificial Intelligence",
address = "Switzerland",

}

TY - CHAP

T1 - A Robust System for High-Quality Reconstruction of 3D Objects from Photographs

AU - Nguyen, H. M.

AU - Wünsche, B. C.

AU - Delmas, P.

AU - Lutteroth, C.

AU - van der Mark, W.

PY - 2014

Y1 - 2014

N2 - Image-based modeling is rapidly increasing in popularity since cameras are very affordable, widely available, and have a wide image acquisition range suitable for objects of vastly different size. In this chapter we describe a novel image-based modeling system, which produces high-quality 3D content automatically from a collection of unconstrained and uncalibrated 2D images. The system estimates camera parameters and a 3D scene geometry using Structure-from-Motion (SfM) and Bundle Adjustment techniques. The point cloud density of 3D scene components is enhanced by exploiting silhouette information of the scene. This hybrid approach dramatically improves the reconstruction of objects with few visual features. A high quality texture is created by parameterizing the reconstructed objects using a segmentation and charting approach, which also works for objects which are not homeomorphic to a sphere. The resulting parameter space contains one chart for each surface segment. A texture map is created by back projecting the best fitting input images onto each surface segment, and smoothly fusing them together over the corresponding chart by using graph-cut techniques. Our evaluation shows that our system is capable of reconstructing a wide range of objects in both indoor and outdoor environments.

AB - Image-based modeling is rapidly increasing in popularity since cameras are very affordable, widely available, and have a wide image acquisition range suitable for objects of vastly different size. In this chapter we describe a novel image-based modeling system, which produces high-quality 3D content automatically from a collection of unconstrained and uncalibrated 2D images. The system estimates camera parameters and a 3D scene geometry using Structure-from-Motion (SfM) and Bundle Adjustment techniques. The point cloud density of 3D scene components is enhanced by exploiting silhouette information of the scene. This hybrid approach dramatically improves the reconstruction of objects with few visual features. A high quality texture is created by parameterizing the reconstructed objects using a segmentation and charting approach, which also works for objects which are not homeomorphic to a sphere. The resulting parameter space contains one chart for each surface segment. A texture map is created by back projecting the best fitting input images onto each surface segment, and smoothly fusing them together over the corresponding chart by using graph-cut techniques. Our evaluation shows that our system is capable of reconstructing a wide range of objects in both indoor and outdoor environments.

UR - http://dx.doi.org/10.1007/978-3-319-06883-1_1

U2 - 10.1007/978-3-319-06883-1_1

DO - 10.1007/978-3-319-06883-1_1

M3 - Other chapter contribution

SN - 9783319068824

T3 - Studies in Computational Intelligence

SP - 3

EP - 15

BT - Issues and Challenges in Artificial Intelligence

A2 - Hippe, Z. S.

A2 - Kulikowski, J. L.

A2 - Mroczek, T.

A2 - Wtorek, J.

PB - Springer International Publishing

ER -