The identification of a parallelogram in the image plane, along with the knowledge of camera parameters, allows metric properties to be recovered. This paper presents a methodology capable of estimating a rectifying homography for images of planar objects without taking any measurement on the world plane. The rectified image will have only an overall scale ambiguity. The method was implemented to create a photogrammetric package for the estimation of the proposed projective transformation. This package was also extended to compute homographic transformations with standard techniques, such as ground control points, rectangles with a known length ratio, and squares. Examples are presented using synthetic and real data acquired for different purposes, including HDR and panoramic photography. Finally, a practical test with two photogrammetric staffs was carried out to check the accuracy of the procedure, starting from the same quantities measured with an optical level.

Date Published: 21 June 2011
PDF: 12 pages
Proc. SPIE 8085, Videometrics, Range Imaging, and Applications XI, 80850R (21 June 2011); doi: 10.1117/12.890036

Published in SPIE Proceedings Vol. 8085:
Videometrics, Range Imaging, and Applications XI
Fabio Remondino; Mark R. Shortis, Editor(s)