les in
cts.
nage-
| ina
tasks,
'ectly
s and
d the
ently
ity to
ews.
e for
y are
both
WS;
trast
and
costs
elly,
ned
ime-
less
data
ined
EM
ique
isity
etry.
1 an
oto-
tion
into
inge
zM-
;, et
are
| of
b. systems designed for specific applications, optimised for
performance but with limited functionality.
Examples of these instruments are the Leica DCCS for
semi-automatic aerotriangulation and Autometric Pegasus
for DEM generation and editing.
Stripped versions of the complete workstations (a) are
covered under this category too;
c. pc-systems designed for specific applications. These are
characterised by somewhat lower accuracy and slightly less
speed, but these realise a greatly reduced cost.
Examples are the R-Wel DMS and Leica DVP for mapping.
4. OUR USE OF DIGITAL PHOTOGRAMMETRY
IN THE PAST, PRESENT AND FUTURE
This chapter describes various projects in the field of digital
photogrammetry carried out the last decade.
Making a subdivision into past, present and future activities is
difficult since the projects are not so well divided in time.
Nevertheless the next set up is made.
4.1 Experiences in the past
Five to ten years ago digital photogrammetry was expected to
be developing into a valid tool for generating cheap and high
dense DEM's and orthophotos. Next to this, the processes of
aerotriangulation could be accelerated by using (semi-)auto-
matic techniques, and mapping could possibly be automated by
using knowledge based systems.
4.1.1 Semi-automatic aerotriangulation The first tests in the
early nineties were focused on semi-automatic aero-
triangulation. The tests on the DCCS of Leica were promising
(Han, 1992), resulting in the purchase of the system in 1991.
This first step into digital photogrammetry was only a small
step: the data capture followed after the triangulation was still
carried out on analytical plotters.
Up to now, we make use of the DCCS. The speed of carrying
out the aerotriangulation is not as promising as we expected,
but comparable to the performance on an analytical instrument.
Moreover, the performance on difficult areas like bare beaches
was poor: matching results were unsatisfactory or wrong too
often. This made us decide to use analytical instruments instead
in these areas.
4.1.2 Orthophotos Orthophotos were expected to be a
supplementary product next to the standard geographical
databases for high ways, rivers and coasts. These orthophotos
could be delivered as digital images, acting as an underground
layer for GIS'ses.
The RWS-users were rather positive about this product, and
they considered it as an useful addition. However, the extra
costs were an insurmountable obstacle then.
Rectified images are used instead in a few cases. Maybe this use
of rectified images is due to the absence of large height
differences in the Dutch terrain: it is not worth much to spent
extra costs for generating orthophotos instead of rectified
images.
4.1.3 DEM generation In 1990, techniques for automatically
generating DEM's from digital images were about to become
operational in a commercial sense (Ackermann, 1991).
945
As described in chapter 2, the beach profile databases need
height information of the Dutch shore. In stead of measuring
profiles manually, digital photogrammetry provides tools for
automatically generating DEM's. These DEM's could possibly
substitute the profile databases in a later stage. In the meantime
profiles could still be extracted from the DEM's by
interpolation.
We carried out benchmark tests, where we compared the
analytically measured profiles to the interpolated profiles out of
the DEM's. Superficially, the results were encouraging, but
after a closer examination, strange errors, sometimes more then
3m, occurred: the DEM's presented non-existing hills, and
some systematical errors on beaches and dunes as well. Most
likely these errors are caused by the lack of texture on the bare
sand beaches and dunes (Wicherson, 1994).
Our RWS-customers concluded the digital photogrammetric
technique not being ready yet for delivering DEM's for their
coastal monitoring.
For further research work with DEM and orthophoto
generation, we purchased the OrthoMAX-module within
TImagine', developed by Autometric. This module supports all
digital photogrammetric activities, except mapping. We use this
module especially for research work; its present performance
and poor flexibility makes it unsuitable in a production
environment.
4.1.4 Knowledge-based systems The generation of high way
and river databases takes about fifty human-years annually. A
slight improvement in efficiency results therefore in a high
saving of money. Since the high way databases contain many
objects with fixed size, colour and relative position, the
mapping procedure could probably be supported by knowledge
based systems.
An external research project, supported by Rijkswaterstaat,
started which had it's subject in updating the high way data-
bases with actual scanned photographs as basis, a high way
database, and a knowledge based computer program. During the
execution of the research work, concentration was put on
recognising edges of high ways with use of the old database,
and measuring and updating the database successively.
This method worked fine, and could easily be translated to
other elongated objects with well-known appearance, like road-
markings (De Gunst and Den Hartog, 1994). However,
developing this method to a user-friendly software package,
would cost too much time compared to the effort: saving of
processing time for a human operator. Moreover, only some
tens of kilometres were constructed in The Netherlands
annually, and thus this method would not deliver the gain
expected.
Another important aspect is that it is not efficient to use digital
photogrammetric workstations for mapping at the moment.
4.2 Present experiences and developments
Our first experiences with digital photogrammetry were not
very encouraging: except the test and purchase of the DCCS,
the tests didn't work out successful so that digital
photogrammetry wasn't implemented in our production
environment yet.
Further research and development of new products is technique-
driven: asking ourselves how the benefits of digital photo-
grammetry and other new techniques could be applied
optimally.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
