Share Email Print
cover

Proceedings Paper

Line correspondence in aerial images using stochastic relaxation and minimum description length criterion
Author(s): Peter Axelsson
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

The problem of line correspondence in multi images is approached using a clustering procedure in object space, thus being independent of the number of images used. Lines are located in the images and by applying object depending restrictions the main directions of the object, corresponding to the vanishing points in the images, can be found. Lines co-linear with the vanishing directions are projected on cluster planes, having the main directions as surface normals. A possible 3D line will then exist where at least two lines intersect in a cluster plane. By accumulating the cluster plane information to the cluster plane axis lines co-planar with each other will be accumulated to each other. For finding the correct maxima on the cluster axis the method of simulated annealing is applied. By using the minimum description length criterion for computing the weights, different types of information are combined in a common framework. When the (sub)optimum solution of the minimizing function is found the line correspondences are directly derived from the found maxima. Results of a building covered by four aerial images are presented.

Paper Details

Date Published: 17 August 1994
PDF: 8 pages
Proc. SPIE 2357, ISPRS Commission III Symposium: Spatial Information from Digital Photogrammetry and Computer Vision, (17 August 1994); doi: 10.1117/12.182915
Show Author Affiliations
Peter Axelsson, Royal Institute of Technology (Sweden)


Published in SPIE Proceedings Vol. 2357:
ISPRS Commission III Symposium: Spatial Information from Digital Photogrammetry and Computer Vision
Heinrich Ebner; Christian Heipke; Konrad Eder, Editor(s)

© SPIE. Terms of Use
Back to Top