ë
*
SAP Xu []1/]] rp:
ct
-
[7
0 a
nea-
sing
ond
n be
1 of
nally
y of
war-
it is
Jlici-
ess.
is a
atric
nap-
tion
jlot-
sta-
'eci-
lase
ym-
the
ints
fer-
ised
oto-
ints
ore
ma-
nali-
ovi-
e.g.
bilinear interpolation). Using automatic matching pro-
cedures a precision of about O.1 pixel can be achieved
on well-defined points. With a pixel size between 10
and 20 um the (geometric) precision of the photogram-
metric scanner is the limiting factor for the precision of
the measurements when matching is applied.
The precision of plotting is dependent on the appea-
rance of the feature in the photograph which varies
with its shape and properties such as reflectance. De-
pending on the characteristics of an imaged ‘feature, a
measuring precision between 10 and 20 um can be
achieved (Timmerman, 1982). Using digital images
plotting can be performed with a standard deviation in
the order of 1 pixel (pixel size between 10 x 10 ym?
and 20 x 20 um?; see section 4.1).
In case of plotting it is expected that the quality of the
measurements in stereo models is somewhat better
than in single photographs due to an improvement of
the interpretation of the imagery.
Although we have specified the measuring precision at
photo scale (i.e. in the photograph), the precision that
can be obtained by photogrammetric techniques de-
pends only partly on the photo scale. The actual preci-
sion is more dependent on the correct interpretation of
the contents of the photographs. The interpretation is
improved considerably by using colour photography.
Systematic errors due to interpretation can be charac-
terised by a standard deviation in the order of 5 to
10 cm at terrain scale.
To safeguard a good interpretation a factor 3 is chosen
for the magnification from photo scale to mapping
scale. This results in the use of two photo scales.
Photo scale 1:3000 is used for built-up areas and
1:6000 for rural areas. With this approach the precisi-
on obtained is somewhat better than the criteria for
the base-map.
The measuring precision (at photo scale) is somewhat
better for the 1:6000 photography. The image quality
of the 1:3000 photography could be slightly degraded
as a result of image motion, if not (fully) compensated
by FMC, or defocus (a 30 cm lens is used at a flying
height of 900 m).
In 1994 a pilot project took place in which a smaller
photo scale was tested for the production of the large-
scale base map of The Netherlands (Mulder, 1994).
The applied photo scales were 1:5000 instead of
1:3000 in built-up areas and 1:12000 instead of
1:6000 in rural areas. This reduction in photo scale
resulted in a reduction of the costs of 30% compared
to the conventional production of the standard GBKN.
The geometric quality of the resulting map meets the
GBKN-requirements. Unfortunately the photo scale
was not the only difference with the production of the
standard GBKN: the map contents was reduced as
well. Because of this and the fact that no comparison
is made with conventional GBKN-production in the
same area, final conclusions on costs, contents or
quality can not be drawn.
4. STANDARDS FOR THE PHOTOGRAM-
METRIC PRODUCTION PROCESS
The photogrammetric production process is split up in
three major steps, one data acquisition step and two
data reduction steps (see figure 2):
1 flightplanning and photoflight;
2 (aero)triangulation and block adjustment;
3 plotting and field completion.
Although the HTW does not prescribe in detail how the
mapping should be performed, it sets standards and
guidelines in order to warrant the quality of the resul-
ting digital map. We will discuss the main standards
for each step.
objectives
limiting conditions
flightplanning
data acquisition
flight (GPS)
Y
(scanned) aerial photographs
Wa control points
GPS positions
triangulation
blockadjustment
coordinates of
tiepoints
Y
plotting / map revision
field completion
connection of point flelds
data reduction 1
data reduction 2 L—»
Y
Figure 2: The process of photogrammetric mapping
The quality of the result of the photogrammetric pro-
duction process is primarily warranted through the
standardization of the process. This standardization
has not been built from scratch. The (former) photo-
grammetric department of the cadastral agency and
other institutions in The Netherlands issued guidelines
for their photogrammetric mapping, but sometimes
these guidelines differ considerably from one another.
The guidelines presented in the HTW are a unification
of the existing guidelines.
A major aspect of the standardization is the introducti-
on of quality control using a testing procedure in every
step of the process. Although the criteria (tolerances)
used in these procedures are quite strict, they are cho-
sen in such a way that they can be met in production
environments.
For the statistical testing of the block adjustment a
level of significance of O.196 is used for the one-di-
mensional test or w-test. Because the so-called B-me-
903
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
