International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 5. Hakodate 1998
CAD-BASED PHOTOGRAMMETRY FOR 3-D RECONSTRUCTION OF LARGE OBJECTS
Thomas Luhmann
Professor and Head of the
Institute for Applied Photogrammetry and Geoinformatics
University of Applied Sciences Oldenburg
Ofener Str. 16, D-26121 Oldenburg
E-mail: lunmann@fh-oldenburg.de
GERMANY
Commission V, Working Group V/2
KEY WORDS: multi-image photogrammetry, image scanning, image database, CAD-based photogrammetry
ABSTRACT
The photogrammetric reconstruction of large buildings is normally characterized by difficult environmental conditions,
complex image configurations and large numbers of images. In this paper a hybrid photogrammetric system is
presented which consists of well-proven analogue imaging devices in combination with a low-cost image scanner and a
CAD-integrated digital multi-image compiling system. Practical results of large engineering projects are used to discuss
experiences and system performance.
1 INTRODUCTION
1.1 Objectives
The photogrammetric reconstruction of large engineering
objects such as manufacturing plants, power stations,
buildings or bridges is not a new application. Various
examples have been published showing the successful
use of photogrammetry for the measurement of complex
environments (e.g. Kotowski et.al. 1989). However, it is
stil an ambitious engineering task due to the complexity
and structure of many objects, and the general
technological and economical situation.
This report presents practical experiences using a hybrid
low-cost photogrammetric system. The system has been
applied to a number of projects which can be
characterized as follows:
recording of large objects
processing of natural, non-targeted object structures
medium accuracy requirements
hidden object areas
difficult environmental conditions
results processed as 3-D CAD data
The spatial dimension of the objects varies between
approx. 10m to 200m. The required accuracy is about
1:5.000 to 1:20.000 of the maximum object dimension.
The object surfaces can not be marked by artificial targets
and image processing is mostly performed manually.
Due to practical conditions and the complexity of many
objects, variable imaging configurations are required. The
photogrammetric network is always designed as a
convergent multi-image configuration whereby local
restrictions often lead to non-optimal ray intersections. As
“ Modified version of (Luhmann 1997)
236
a result, multi-image orientation is performed by bundle
adjustment. The datum problem is normally solved by
locally provided control points. The configuration of
control points and camera stations must enable
simultaneous camera calibration.
analogue digital :
(photographic) (optoelectronical) |
imaging
development digitization
Y
analogue digital :
(visual) (automated)
processing processing |
Figure 1: Principles of photogrammetric
data acquisition and processing
1.2 Solutions
The highlighted sequence in Figure 1 describes the
hybrid photogrammetric process used in the following
applications. It is based on analogue photogrammetric
image acquisition. Using low-cost desktop image
scanners enlarged paper prints can be digitized with
sufficient geometric resolution and accuracy. Finally, the
photogrammetric image evaluation is performed with
interactive digital processing systems that are integrated
into a CAD environment. This method combines the
advantages of a well-proven imaging technology with
those of digital image and CAD processing.
2.1 Image
All projects
metric carr
150mm us
resolution i
negatives |
18cm (scali
A simple c
used for im
radiometric
50um) do !
system. Hc
yields to al
original fil
correction (
accuracies
Recently a
has been U
Scandet is
strong noi
adjustment
fulfil the rec
2.2 Image
Large and
images) re
2.3 Photc
The intera
(Benning
199???) a
have been
Systems s
developed.
