h
ut
he
20
hy
to
1a
ed.
30
the
01)
be
-10
for
ng,
cen
'ans
‘the
2 83
of
num
rfect
-flat
nage
nials
d be
ising
ibber
high
area
ional
ne, it
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
3.5 Synthetic Stereo-Image creation
The geological and geomorphological characters of the area
could be better analysed using the stereo instead the mono
vision. Traditionally the stereoscopic vision is based on a
couple of aerial photographs or satellite images following the
geometric characteristics of acquisition. In order to obtain the
stereoscopic viewing from a single QuickBird imagery, a
second synthetic stereo-image on a PC must be created. The
PCI Geomatica 8.0 software was utilized to introduce an
artificial parallax to the orthorectified image in order to create
the new synthetic scene. Following the DEM, the synthetic
image was created analytically by the introduction of parallax
values directly proportional to the ground elevation (Batson et
al., 1976). The value of the artificial parallax (DP) for a single
pixel on the image is expressed by the following formula:
DP - Dh*K (1)
where Dh = elevation of the pixel above the minimum
ground elevation
K = constant value determining the strength of the
stereoscopic vision
The STE (STereo image pair from Elevation data) module, of
PCI Geomatica 8.0, based on the above relation and it works
with single band images. The following are the STE module's
parameters:
- Stereoscopic Factor: defines the strength of the artificial
parallax introduced on the orthorectified image. The
stereoscopic factor is proportional to the vertical
exaggeration. The more shifts introduced to the image, the
more vertical exaggeration the — stereomate has
(Methakullachat, 1994). The K value of the relation is the
mutual one of the Stereoscopic Factor. The Stereoscopic
Factor utilized for the production of the QuickBird
synthetic imagery was 2.92;
- Elevation Step Size: defines, in meters, the variation of
ground elevation correspondent to unity. The correct
creation of a synthetic image depends on right specification
of this value. In this work it was defined equal to |.
- Port settings: indicates the algorithm of resampling used in
the creation of the new synthetic imagery (the nearest
neighbour was selected).
3.6 Stereo-photogrammetry
After the creation of the synthetic imagery for the stereo vision,
the choice was among the hard print of the stereopair to be
observed under a mirror stereoscope, the production of an
anaglyphic image, and the orientation by means of a
photogrammetric workstation. This last was taken on, and the
Leica Photogrammetric Suite (LPS) Module of Erdas Imagine
8.7 software was utilized. For the interior orientation the *Non-
Metric Camera" Model has been selected. The exact QuickBird
focal length by DigitalGlobe communication (2004) was made
known. The pixel size (PX) was determined by the following
formula:
HS: RG = FL: PX (2)
where — HS — orbital altitude (450000 m)
RG = ground resolution (0.7 m)
FL — focal length (8.836202 m)
The calculated pixel size is 13.745 micron.
The exterior orientation was realized by using 50 GCPs from
the DGPS measurements. 198 Tie Points automatically
generated by the software and accurately edited integrated the
relative orientation.
Taking into account the DGPS data accuracy, the triangulation
was performed with a standard deviation of 1 meter for the
coordinates X, Y and Z of each GCP; the RMSE results 0.9785
pixels (approximately 0.68 meters).
4. RESULTS
The described methods allowed the digital stereo-interpretation
of the QuickBird satellite imagery. Basing on archaeological
priorities, the geological map at a scale of 1:10,000 of the Ain
Hofra Wadi (Figure 4) and the Belghadir Wadi were first
created.
21*51'50" 215230"
Legend
crm ea
I] Compare
Hydrog apıy
o Spr
tana
+ Beddhg strke 214 dp
1 Faut
Geomorphology
IN S 360
Wadi Ain Hofra
>
y
32
Figure 4. Geological map of the Ain Hofra Wadi
The whole necropolis of Shahhat expands basically on two
types of rocks, limestone with nummulites at the bottom and
marly seaweed limestone at the top.
In the Ain Hofra area, the one actually under archaeological
digging, the geomorphology is characterised by the presence of
paleo-springs just below the present ground level; about 15
meters below, an active spring-rising, coming from karstic
phenomenon and/or from jointing, is present. The
geomorphological evolution, that is a consequence of climatic
factors and gravity, has actively contributed to the erosion and
the engraving of the canyon of Ain Hofra and to the selective
erosion of the outcropping bedding. Rock falls have been
noticed on the upper part of the canyon, just under the present
archaeological diggings and on the western side deep-seated
gravitational deformations have been observed.
From a technical point of view, the present work gave several
remarks. The accuracy of the orthorectification, is basically
related to the following factors: a) input data type (a raw
"basic" data, which on the other hand is geographically bigger,
much more expensive and it requires a proper software to be
processed, can be better orthorectified; Toutin and Cheng,
2002); b) the small scale of the topography; c) the accuracy of
the DGPS measurements. The first factor is essentially a matter
of money, the second and the third depend on data availability;
for instance, a better result on DGPS measurements would have
been achievable with either reference points of known
1103
