Advanced

Practical Robust Two-View Translation Estimation

Fredriksson, Johan LU ; Larsson, Viktor LU and Olsson, Carl LU (2015) 28th Conference on Computer Vision and Pattern Recognition, 2015 In 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) p.2684-2690
Abstract
Outliers pose a problem in all real structure from motion

systems. Due to the use of automatic matching methods

one has to expect that a (sometimes very large) portion of

the detected correspondences can be incorrect. In this paper

we propose a method that estimates the relative translation

between two cameras and simultaneously maximizes

the number of inlier correspondences.

Traditionally, outlier removal tasks have been addressed

using RANSAC approaches. However, these are random in

nature and offer no guarantees of finding a good solution.

If the amount of mismatches is large, the approach becomes

costly because of the need to evaluate a... (More)
Outliers pose a problem in all real structure from motion

systems. Due to the use of automatic matching methods

one has to expect that a (sometimes very large) portion of

the detected correspondences can be incorrect. In this paper

we propose a method that estimates the relative translation

between two cameras and simultaneously maximizes

the number of inlier correspondences.

Traditionally, outlier removal tasks have been addressed

using RANSAC approaches. However, these are random in

nature and offer no guarantees of finding a good solution.

If the amount of mismatches is large, the approach becomes

costly because of the need to evaluate a large number of

random samples. In contrast, our approach is based on the

branch and bound methodology which guarantees that an

optimal solution will be found. While most optimal methods

trade speed for optimality, the proposed algorithm has competitive

running times on problem sizes well beyond what is

common in practice. Experiments on both real and synthetic

data show that the method outperforms state-of-the-art alternatives,

including RANSAC, in terms of solution quality.

In addition, the approach is shown to be faster than

RANSAC in settings with a large amount of outliers. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Two-view estimation, computer vision, optimization
in
2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
pages
7 pages
publisher
Computer Vision Foundation
conference name
28th Conference on Computer Vision and Pattern Recognition, 2015
external identifiers
  • Scopus:84959249513
DOI
10.1109/CVPR.2015.7298884
language
English
LU publication?
yes
id
a68dd0a6-edd3-4f92-af85-683396b0c302 (old id 5276657)
alternative location
http://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Fredriksson_Practical_Robust_Two-View_2015_CVPR_paper.pdf
date added to LUP
2015-07-07 17:19:04
date last changed
2016-10-23 04:35:01
@misc{a68dd0a6-edd3-4f92-af85-683396b0c302,
  abstract     = {Outliers pose a problem in all real structure from motion<br/><br>
systems. Due to the use of automatic matching methods<br/><br>
one has to expect that a (sometimes very large) portion of<br/><br>
the detected correspondences can be incorrect. In this paper<br/><br>
we propose a method that estimates the relative translation<br/><br>
between two cameras and simultaneously maximizes<br/><br>
the number of inlier correspondences.<br/><br>
Traditionally, outlier removal tasks have been addressed<br/><br>
using RANSAC approaches. However, these are random in<br/><br>
nature and offer no guarantees of finding a good solution.<br/><br>
If the amount of mismatches is large, the approach becomes<br/><br>
costly because of the need to evaluate a large number of<br/><br>
random samples. In contrast, our approach is based on the<br/><br>
branch and bound methodology which guarantees that an<br/><br>
optimal solution will be found. While most optimal methods<br/><br>
trade speed for optimality, the proposed algorithm has competitive<br/><br>
running times on problem sizes well beyond what is<br/><br>
common in practice. Experiments on both real and synthetic<br/><br>
data show that the method outperforms state-of-the-art alternatives,<br/><br>
including RANSAC, in terms of solution quality.<br/><br>
In addition, the approach is shown to be faster than<br/><br>
RANSAC in settings with a large amount of outliers.},
  author       = {Fredriksson, Johan and Larsson, Viktor and Olsson, Carl},
  keyword      = {Two-view estimation,computer vision,optimization},
  language     = {eng},
  pages        = {2684--2690},
  publisher    = {ARRAY(0x9c48618)},
  series       = {2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  title        = {Practical Robust Two-View Translation Estimation},
  url          = {http://dx.doi.org/10.1109/CVPR.2015.7298884},
  year         = {2015},
}