MERA
ping, a
ie large
use of a
: The
llowing
bundle
ation of
he high
nd well
cted by
images.
| points
can be
Icreases
iore tie,
stereo-
n result
rections
and the
e aerial
'oposed
Marger.
. Aerial
ed by 4
d scale
proved
rincipal
original
gv" the
. points
recious
esult of
ige loss
original
are not
results
stection
nd the
| over-
-variant
; small-
terrain
nmetric
bundle
and
for the
Metric cameras on the other hand offer the following
advantages :
-Self-calibration is not required since the metric cameras are
pre-calibrated, have fiducial marks and minimum image
distortions. As a result, aerotriangulation by independent
models can be applied. The method of independent models :
* js not as sensitive to flat terrain and vertical photography.
* uses the perspective centers as pass points, which strengthen
the connection of the models.
* does not require dense control and image points.
-Less photographs are needed to cover the tidal areas of interest
with the required scale resulting in a great reduction at the
model connections, the unknown parameters and the control
points.
-They have high image quality compared with the enlargements
and therefore offer better measurement accuracy, easier
selection of pass and tie points and reduced numbers of small
gross errors.
-The acquisition of the photographs is controlled much better,
and the areas of interest are covered properly.
When the cost of covering the same tidal areas with metric and
non-metric cameras was compared, it was surprisingly found
that for having in hand the desired final photographic product it
is more expensive to use non-metric cameras than metric ones.
The reason for this is the high cost of the enlargement of the
small format images. In recent years, the developments of the
color image copiers enable us to enlarge images by digital
scanning rather than by using purely optical means. The laser
scanned copier, unlike the photographic enlarger, does not
introduce significant image distortions (e.g. CLC-200
enlargement accuracy is approximately 15 um (Warner and
Andersen, 1992)). Moreover the enlargements made with the
laser scanned copier are much cheaper, almost 1/10th of the
cost of conventional enlargements. The problem that still
remains, even when a color laser copier is used is the cropping
of the image.
Concluding the preceding discussion it is suggested that the
tidal terrain is mapped using only metric cameras.
5. FROM AERIAL PHOTOGRAPHY TO DTM OF
TIDAL AREAS
Defining the objectives and the area of interest. The DTM
which was based on photogrammetric observations of heights
of points at the sea-bed was generated in order to be used as
ground truthing for testing the accuracy and the general
response of the acoustic methods for ocean mapping tasks.
Because the expected accuracy of the acoustics was quite high
and the investigation of the degree of their sensitivity to small
objects was of great interest, features that show abrupt changes
in height must be selected as the test objects. Since the tidal
areas are mostly flat and featureless the only such objects that
were found at the areas of investigation were three rocks close
to McNamara point in Saint John Harbour (see Figure 5.1). One
of the rocks is small with sizes about 25 m by 54 m and the
other two are bigger, 40 m by 90 m and 60 m by 142 m
respectively.
In order for the DTM produced by photogrammetric methods to
be used as ground truthing it must have an accuracy that is
higher than the most optimistic estimation of the accuracy
achieved by the acoustic methods. It was decided that a DTM
with 25 cm accuracy will be sufficient for the test.
21
Figure 5.1
The three rocks test tidal area in Saint John Harbour.
Acquisition of the aerial photography. The acquisition of the
aerial color photography was carried out on the 20th of
September 1993 at 10:30 am. A metric camera Wild RC-10 was
used for the acquisition of the vertical photography at a 1/4000
scale. The whole Saint John Harbour area was covered with 35
photographs having an over-lap of 6096 and side-lap of 30%.
The tidal area of interest (the three rocks) was also covered by
five photographs with 8096 over-lap.
Collection of the DTM primary data. The static method
(photogrammetric observations of spot heights) was used for
measuring of the DTM primary data. The photogrammetrically
observed spot heights were collected using composite sampling.
The sea-bed is very flat with the exception of the three rocks
and the features of interest are mainly these rocks. The smaller
rock was covered by photogrammetric observations of spot
heights at a grid interval of 40 cm. The grid interval for the
other two rocks was 80 cm for one, and 1 m for the other. The
rest of the area was covered by a grid with an interval of 10 m.
Since the rocks show abrupt changes in elevation and contain a
number of small and big stones, selected spot heights were
photogrammetrically observed in places where they were
needed in order to represent the surface better (see Table 5.1).
Table 5.1
Number of spot height observations according to the type of sampling and the
sampled feature.
40 cm 80 cm 1m 10m selective total
rock 3714 708 4422
a 950 6693
712 934
2881 637 51
3714 3222 5743 2881 2997 18567
DTM generation. The photogrammetrically collected DTM
primary data were processed in a Sun-Sparc station using the
CARIS GIS. The digital terrain model was generated using the
TIN method. Contour maps, perspective views, shaded relief
representations, superimpositions of digital images of the area
with contours, superimposition of perspective views with
contours and colour-coded representation of height, aspect and
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996