Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Practical global optimization for multiview geometry

Agarwal, S ; Chandraker, MK ; Kahl, Fredrik LU ; Kriegman, D and Belongie, S (2006) In Lecture Notes in Computer Science 3951(Pt 1: Proceedings). p.592-605
Abstract
This paper presents a practical method for finding the provably globally optimal solution to numerous problems in projective geometry including multiview triangulation, camera resectioning and hemography estimation. Unlike traditional methods which may get trapped in local minima due to the non-convex nature of these problems, this approach provides a theoretical guarantee of global optimality. The formulation relies on recent developments in fractional programming and the theory of convex underestimators and allows a unified framework for minimizing the standard L-2-norm of reprojection errors which is optimal under Gaussian noise as well as the more robust L-1-norm which is less sensitive to outliers. The efficacy of our algorithm is... (More)
This paper presents a practical method for finding the provably globally optimal solution to numerous problems in projective geometry including multiview triangulation, camera resectioning and hemography estimation. Unlike traditional methods which may get trapped in local minima due to the non-convex nature of these problems, this approach provides a theoretical guarantee of global optimality. The formulation relies on recent developments in fractional programming and the theory of convex underestimators and allows a unified framework for minimizing the standard L-2-norm of reprojection errors which is optimal under Gaussian noise as well as the more robust L-1-norm which is less sensitive to outliers. The efficacy of our algorithm is empirically demonstrated by good performance on experiments for both synthetic and real data. An open source MATLAB toolbox that implements the algorithm is also made available to facilitate further research. (Less)
Please use this url to cite or link to this publication:
author
; ; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
in
Lecture Notes in Computer Science
volume
3951
issue
Pt 1: Proceedings
pages
592 - 605
publisher
Springer
external identifiers
  • wos:000237552900046
  • scopus:33745866266
ISSN
1611-3349
language
English
LU publication?
yes
id
6b4832e5-2faf-4da3-b765-b484c2c3573b (old id 405669)
alternative location
http://vision.ucsd.edu/kriegman-grp/papers/eccv06b.pdf
date added to LUP
2016-04-01 12:02:38
date last changed
2021-10-06 05:39:43
@article{6b4832e5-2faf-4da3-b765-b484c2c3573b,
  abstract     = {This paper presents a practical method for finding the provably globally optimal solution to numerous problems in projective geometry including multiview triangulation, camera resectioning and hemography estimation. Unlike traditional methods which may get trapped in local minima due to the non-convex nature of these problems, this approach provides a theoretical guarantee of global optimality. The formulation relies on recent developments in fractional programming and the theory of convex underestimators and allows a unified framework for minimizing the standard L-2-norm of reprojection errors which is optimal under Gaussian noise as well as the more robust L-1-norm which is less sensitive to outliers. The efficacy of our algorithm is empirically demonstrated by good performance on experiments for both synthetic and real data. An open source MATLAB toolbox that implements the algorithm is also made available to facilitate further research.},
  author       = {Agarwal, S and Chandraker, MK and Kahl, Fredrik and Kriegman, D and Belongie, S},
  issn         = {1611-3349},
  language     = {eng},
  number       = {Pt 1: Proceedings},
  pages        = {592--605},
  publisher    = {Springer},
  series       = {Lecture Notes in Computer Science},
  title        = {Practical global optimization for multiview geometry},
  url          = {http://vision.ucsd.edu/kriegman-grp/papers/eccv06b.pdf},
  volume       = {3951},
  year         = {2006},
}