contrast to the white background.
Fig.(1) illustrates the roof model and also shows the
arrangement of the ground control points.
Fig.(1) The Structural Model with Control Points
3. COORDINATION OF OBJECT SPACE CONTROL
As shown in Fig.(1), 18 control points with different elevations
were placed around the model with 3 more located inside the
model to improve the geometry of the photogrammetric
network.
Similar to the wires, the shape, size, and colour of the control
targets were investigated. Either square or circular targets
with a black and white combination for contrast were found
to be the most suitable for all cases. Their size of 1 cm
accommodates the range of photo scales used.
Since the accuracy of the control points directly influences
the photogrammetric results, a control survey of high
accuracy was carried out using an Electronic Coordinate
Determination System. This near real-time 3D coordinate
measuring system consists of two one-second electronic
theodolites ( Kern E2 ) interfaced with an IBM PC computer.
A 2 metre invar bar with a calibrated length of 2.10053 m was
used to provide the scale of the system. As illustrated in
Fig.(2), a near ideal configuration was selected for the control
survey.
The observation procedure is similar to traditional intersection
with horizontal directions and zenith distances observed for
each point to be coordinated. The observed values are
directly fed into the computer for further processing,
therefore, no hand-booking is required. Immediately after
each observation set, a bundle adjustment was carried out to
acquire 3D coordinates for all the unknown points as well as
pertinent accuracy information.
All horizontal directions and zenith distances were observed
in two sets. The average standard deviation for the final
coordinates of the control points were 0.03 mm, 0.025 mm,
0.02 mm in X, Y, and Z directions, respectively.
Invar Bar
Model
Y
A
|
|
|
|
|
|
|
Kern| E2 Kern E2
ds — »X
Fig.(2) Control Survey Configuration
4.IMAGE CONFIGURATION
Generally speaking, deformation monitoring with photo-
grammetric techniques is based on taking some photographs
before and after the deformation from appropriately located
camera stations ( Erlandson, 1975 ). After the data
evaluation, two sets of 3D coordinates of the detail points
(intersections in our case), which represent the roof surface,
can be obtained for both epochs. The differences between
the two sets of coordinates represent the deformation
information.
Since geometric configuration is one of the main factors that
influence the final accuracy, sophisticated methods have
been designed especially for the purpose of network design
(C.S.Fraser, 1984). In our case of study, a multi-station
convergent photogrammetric network was established to
monitor the deformation.
Taking into account the economical factor, as well as the
uniform accuracy desired for the three directions X, Y, and Z,
three camera stations were selected with 100% overlap for
any pair of convergent photos. Fig.(3) illustrates the camera
AY
Station(3)
135... 135. Model
SON
O O
© O > X
Station (1) Station (2)
Fig.(3) Camera Locations Lay out
location lay out.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996
Photographs
deformation \
with the same
of the imagin:
emulsion for
the digital ca
two options \
object distanc
each camera
change the c
From a pract
The second
expected she
and sharp im
distance was
scale of 1:90
digital camer:
the tension o
hanging wire:
The non-metri
with 28 mm
format, f-stop
time adaptor €
placed in the
medium was
While captur
which repres
mm, to maxim
the laboraton
A Fujix DS-
utilized. It is
specifically «
purposes. It
camera bece
teristics as ct
medium.
Its focal lengtk
power zoom
exposure tir
element is a
image senso
with a pixel si
a IM-85 Ima
images storec
the computer
SD-PC (Dig
software. The
pixels with a
For image c:
chosen to ok
the maximun
time ( t=1/4 s
In addition to
digital ), a thirc
hard copy im
this task. Th«
millimetre) ar
