International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
This paper investigates the contribution of laser scanning into
the production of large scale orthophotographs. A case study
which is presented in this context uses data collected from a
relatively small 15" century Byzantine church comprising a
variety of surfaces. Two typically different types of surfaces
have been chosen to illustrate the differences in the
improvement of the final end products. In addition to
conventional geodetic and photogrammetric data acquisition
using analog cameras, laser scanning with a Cyrax 2500 laser
scanner was carried out. A comparison is performed between
orthophotographs produced using laser scanned data and
conventional surface descriptions.
2. DATA COLLECTION
The case study presented in this paper involves a 15" century
stone-built Byzantine church (Figure 1) located in the island of
Tilos, in the southeast Aegean. The church belongs to the
complex of the St. Panteleimon Monastery and was built on top
of the ruins of an ancient temple dedicated to Apollo. As seen in
Figure 1, the church is surrounded by high walls which were
constructed very close to the monument for protection. The
non-built area between the church and the walls is limited thus
posing many difficulties in the field procedures during data
acquisition. The plan view of the church is shown in Figure 2.
Figure 1. South-eastern view of the church
Photogrammetric and laser scanner data collection was
performed for most external surfaces of the church. However,
data only from two parts of the monument are shown here:
e the data collected from a tiled roof at the north-western part of
the church, which will be referred to as “data set I" in the
remainder of this paper
e the data collected from the eastern part of the church, which
will be referred to as “data set 11”.
The chosen parts comprise complex surfaces with intense relief,
thus highlighting the advantages of integrating both types of
data.
Specifically, the image data acquisition of the north-western
part was performed with the semi-metric camera Rolleiflex
6006, format 5.5x5.5 cm? and c-40mm. Four photographs were
taken which produced two stereopairs at a scale of about 1:80.
The geometry of the stereopairs was adverse, due to the narrow
space available and the existing obstacles in the area. The first
pair (camera stations Cl and C2 in Figure 2) had an
unfavourable ratio B/H=1:8, while the second pair which had a
ratio B/H=1:2 (camera stations C3 and C4) was formed by
converging bundles at an angle of approx. 24". The image
468
acquisition for the eastern part was performed with the non-
metric camera Hasselblad C/M 500, format 5.5x5.5 cm? and
c=50mm. A total of four photographs were acquired, which
resulted in two stereo-pairs. The distance of the acquired photos
varied from about 10m for the first two, i.e. approximate photo
scale at 1:200 (camera stations C5 and C6 in Figure 2), to 13m
for the remaining two photos, i.e. approximate photo scale at 1:
260 (camera stations C7 and C8). These pairs had also a small
B/H ratio, especially the second, which had a ratio B/H=1:10,
and in addition to that they did not fully cover stereoscopically
the whole of the right half of the facade. All images were taken
using colour slide film.
The laser scanner data acquisition was performed with a Cyrax
2500 instrument, which was mounted on its tripod during data
capturing. Three scans were required to capture the north-
western part of the monument with point density of 0.020m. The
eastern. part. was captured with two scans from different
locations and at a point density of 0.025m (Figure 2). An
overlapping of about 40% was used to cover undercuts and
hidden zones. The scans resulted to a total of about 1.5 million
points. Figure 2 illustrates the locations for both camera and
laser scanner set ups along with their relevant cone capture.
sconner setup
Figure 2. Plan view of the church, showing scanner set ups and
camera positions (not to scale)
A general requirement for all surveys was a common coordinate
system. A precise network of 16 traverse stations around the
church was established using a Leica TC307 total station,
resulting to an accuracy of better than 4 mm. Using these
network points, about 50 control points were then measured
comprising targets for the photogrammetric restitution and
Cyrax targets for the point cloud registration and
georeferencing. Processing of the laser scan data was performed
in the Cyclone software. The registration of the data was
achieved within an accuracy of about Smm and the
georeferencing within Imm.
3. DATA PROCESSING
The acquired data from both methods were edited separately, so
that the necessary digital surface models (DSM) would be
created, for the orthophoto production of the north (dataset I)
and eastern (dataset II) facade of the church.
3.4 Photogrammetric Procedure
All the photogrammetric works were performed with the digital
workstation Softplotter v.4 of Autometric. The films were
scanned with a resolution of 1600 dpi. The first processing stage
is the completion of the orientations. The parameters of the
calibration of the two cameras were estimated in order to
International
ue
perform inte
relative and
photos each,
using 10 pre
dataset [ and
made for pe
estimation of
of the resultii
The next st
extraction. 1
Irregular Net
best ways of
by two diffe
As it was «
“describe the
approach wa
involved usi
breaklines, s
set I, which
were made:
small numb:
breaklines (v
Due to the
experiment £
contrary, the
of points a
carefully sel
2cm at grout
both DSMs
the objects
difficult, aln
(c.g. the rig
eastern fagac
3.2 Laser §
The process
Cyclone 4.
accomplishe
registration
process of 1
system, de:
orientation,
provides th
methods; th
or using a «
combined r
acquired |
registration
eastern faca
therefore, r
registration
The next st
transformati
system. It i
gco-referenc
Surveying [
process. In |
part was pe
specific inst
less than 71
the eastern
cloud with
coordinates
