Advanced

Structure and Motion from Points, Lines and Conics with Affine Cameras

Kahl, Fredrik LU and Heyden, Anders LU (1998) Computer Vision - ECCV'98 5th European Conference on Computer Vision In [Host publication title missing] 1. p.327-341
Abstract
We present an integrated approach that solves the structure and motion problem for affine cameras. Given images of corresponding points, lines and conics in any number of views, a reconstruction of the scene structure and the camera motion is calculated, up to an affine transformation. Starting with three views, two novel concepts are introduced. The first one is a quasi-tensor consisting of 20 components and the second one is another quasi-tensor consisting of 12 components. These tensors describe the viewing geometry for three views taken by an affine camera. It is shown how correspondences of points, lines and conics can be used to constrain the tensor components. A set of affine camera matrices compatible with the quasi-tensors can... (More)
We present an integrated approach that solves the structure and motion problem for affine cameras. Given images of corresponding points, lines and conics in any number of views, a reconstruction of the scene structure and the camera motion is calculated, up to an affine transformation. Starting with three views, two novel concepts are introduced. The first one is a quasi-tensor consisting of 20 components and the second one is another quasi-tensor consisting of 12 components. These tensors describe the viewing geometry for three views taken by an affine camera. It is shown how correspondences of points, lines and conics can be used to constrain the tensor components. A set of affine camera matrices compatible with the quasi-tensors can easily be calculated from the tensor components. The resulting camera matrices serve as an initial guess in a factorisation method, using points, lines and conics concurrently, generalizing the well-known factorisation method by Tomasi-Kanade (1992). Finally, examples are given that illustrate the developed methods on both simulated and real data (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
cameras, computational geometry, computer vision, constraint theory, image reconstruction, matrix decomposition, motion estimation, tensors
in
[Host publication title missing]
volume
1
pages
327 - 341
publisher
Springer
conference name
Computer Vision - ECCV'98 5th European Conference on Computer Vision
external identifiers
  • Scopus:84957650663
ISBN
3 540 64569 1
language
English
LU publication?
yes
id
a968f78c-f0a6-4137-b502-3be6767b6814 (old id 787363)
date added to LUP
2008-03-31 17:33:27
date last changed
2016-10-13 04:50:38
@misc{a968f78c-f0a6-4137-b502-3be6767b6814,
  abstract     = {We present an integrated approach that solves the structure and motion problem for affine cameras. Given images of corresponding points, lines and conics in any number of views, a reconstruction of the scene structure and the camera motion is calculated, up to an affine transformation. Starting with three views, two novel concepts are introduced. The first one is a quasi-tensor consisting of 20 components and the second one is another quasi-tensor consisting of 12 components. These tensors describe the viewing geometry for three views taken by an affine camera. It is shown how correspondences of points, lines and conics can be used to constrain the tensor components. A set of affine camera matrices compatible with the quasi-tensors can easily be calculated from the tensor components. The resulting camera matrices serve as an initial guess in a factorisation method, using points, lines and conics concurrently, generalizing the well-known factorisation method by Tomasi-Kanade (1992). Finally, examples are given that illustrate the developed methods on both simulated and real data},
  author       = {Kahl, Fredrik and Heyden, Anders},
  isbn         = {3 540 64569 1},
  keyword      = {cameras,computational geometry,computer vision,constraint theory,image reconstruction,matrix decomposition,motion estimation,tensors},
  language     = {eng},
  pages        = {327--341},
  publisher    = {ARRAY(0xaccf6a8)},
  series       = {[Host publication title missing]},
  title        = {Structure and Motion from Points, Lines and Conics with Affine Cameras},
  volume       = {1},
  year         = {1998},
}