3D reconstruction using trinocular geometry

Authors

  • Gabriel Conceição Andrade Universidade do Estado do Rio de Janeiro
  • Germano Monerat Universidade do Estado do Rio de Janeiro
  • Juliana Ventura Universidade do Estado do Rio de Janeiro
  • Francisco De Moura Neto Universidade do Estado do Rio de Janeiro
  • Ricardo Fabbri Universidade do Estado do Rio de Janeiro

Keywords:

Computer vision, Photogrammetry, 3D Reconstruction, Free Software

Abstract

Abstract: Structure from motion is a computer vision problem that seeks obtain tridimentional scenes from a set of images shot from different points of view without having any previously camera configuration knowledge. The main approach consists in building a base reconstruction with two images, followed up by insertion of images one by one until the complete reconstruction. In spite of it’s robustness, there are cases that it’s not possible to make the two-view camera initialization leaving SfM systems not conclude the reconstruction process. Recently it is used three image as geometric base that shows bigger robustness potential by providing better reliability in getting correspondences and cameras being proposed as an alternative when two-view initialization fails. Only recently pratical ones could being made due the improvement of solver thechnnieques. These improvements made relative camera pose estimation more robust and efficient by using three oriented SIFT features. This article exposes the recent advances in trinocular geometry into the open source software openMVG alongside provided experimental results showing it’s practical robustness.

Downloads

Download data is not yet available.

References

Apple ARKit Team. Understanding arkit tracking and detection. WWDC, 2018. URL https://developer.apple.com/videos/play/wwdc2018/610.

Ricardo Fabbri, Timothy Duff, Hongyi Fan, Margaret H Regan, David da Costa de Pinho, Elias Tsigaridas, Charles W Wampler, Jonathan D Hauenstein, Peter J Giblin, Benjamin Kimia, and Tomas Pajdla. Camera pose estimation using first-order curve differential geometry. IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 1–1, 2022.

Olivier Faugeras and Quang-Tuan Luong. The Geometry of Multiple Images. MIT Press, Cambridge, MA, USA, 2001. ISBN 0262062208.

R. I. Hartley and A. Zisserman. Multiple View Geometry in Computer Vision. Cambridge University Press, ISBN: 0521540518, second edition, 2004.

Jonathan D. Hauenstein and Andrew J. Sommese. What is numerical algebraic geometry? Journal of Symbolic Computation, 79: 499–507, 2017. ISSN 0747-7171. doi: ttps://doi.org/10.1016/j.jsc.2016.07.015. URL: https://www.sciencedirect.com/science/article/pii/S0747717116300529. SI: Numerical Algebraic Geometry.

Internet. Ransac. https://en.wikipedia.org/wiki/Random_sample_consensus, August 2023a.

Pierre Moulon, Pascal Monasse, Romuald Perrot, and Renaud Marlet. OpenMVG: Open multiple view geometry. In International Workshop on Reproducible Research in Pattern Recognition, pages 60–74. Springer, 2016.

Johannes Lutz Schonberger¨ and Jan-Michael Frahm. Structure-from-motion revisited. In Conference on Computer Vision and Pattern Recognition (CVPR), 2016.

Johannes Lutz Schonberger, Enliang Zheng, Marc Pollefeys, and Jan-Michael Frahm. Pixelwise view selection for unstructured multi-view stereo. In European Conference on Computer Vision (ECCV), 2016.

Downloads

Published

2024-01-31

How to Cite

Conceição Andrade, G., Monerat, G., Ventura, J., De Moura Neto, F., & Fabbri, R. (2024). 3D reconstruction using trinocular geometry. Revista Interdisciplinar De Pesquisa Em Engenharia, 9(2), 83–90. Retrieved from https://periodicos.unb.br/index.php/ripe/article/view/52293

Issue

Vol. 9 No. 2 (2023): Revista Interdisciplinar de Pesquisa em Engenharia

Section

Artigos

License

Copyright (c) 2024 Revista Interdisciplinar de Pesquisa em Engenharia

Creative Commons License

This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License.

Given the public access policy of the journal, the use of the published texts is free, with the obligation of recognizing the original authorship and the first publication in this journal. The authors of the published contributions are entirely and exclusively responsible for their contents.

1. The authors authorize the publication of the article in this journal.
2. The authors guarantee that the contribution is original, and take full responsibility for its content in case of impugnation by third parties.
3. The authors guarantee that the contribution is not under evaluation in another journal.
4. The authors keep the copyright and convey to the journal the right of first publication, the work being licensed under a Creative Commons Attribution License-BY.
5. The authors are allowed and stimulated to publicize and distribute their work on-line after the publication in the journal.
6. The authors of the approved works authorize the journal to distribute their content, after publication, for reproduction in content indexes, virtual libraries and similars.
7. The editors reserve the right to make adjustments to the text and to adequate the article to the editorial rules of the journal.