A REVIEW OF CLOSE-RANGE ENGINEERING PHOTOGRAMMETRY 65
Fic. 13. The underside of the box girder
bridge model, showing some of the pre-
marked points whose relative movements are
to be determined.
which were attached to the camera and three
auxiliary flash units which were mounted
around the box. The use of a short flash dura-
tion overcame any problems which might
have arisen from camera movement.
The photography was observed on a Hilger
and Watts stereocomparator. Relative and ab-
solute orientations were computed analyti-
cally. Accuracies in the horizontal plane were
good but movements in depth were deter-
mined less satisfactorily (average root mean
square error at height control was 0.34 mm).
However, their determination was not so im-
portant in these tests. Scale control at the base
of the box and triangulation of the control
points which were provided would have im-
proved this accuracy.
A second deformation measurement proj-
ect is now in progress. The author is indebted
to Mr. P.J. Scott of University College Lon-
don for the note which follows.
"A 1:12 scale model ofa bifurcated box gird-
er bridge (Figure 12) is to be load-tested at
Imperial College of Science and Technology.
Measurements will only be required in the
localised areas where the failure occurs but
their location will not be known until the
failure has taken place. Close-range photo-
grammetry thus provides the means of record-
ing all points and of providing co-ordinates for
only the required few when their location is
known. The model bridge carries approxi-
mately 4000 premarked targets (Figure 13).
Relative movement between them is to be de-
termined to better than 0.3 mm. This require-
ment has necessitated the use ofa small object
distance of 2.3 m. The shape of the bridge
dictates the large depth of field of 1.8 m to
3.4 m. The majority of the related research is
thus centered around the variation of the
principal distance with object distance
throughout the field. A photographic calibra-
tion method has been devised which evalu-
ates this variation as well as the change in
principal distance with radial distance. In
order to calculate correction terms for ob-
served co-ordinates, the approximate dis-
tance of each target will have to be deter-
mined.
Control points on the bridge will take the
form of marks on plumb wires, each 1 mm in
diameter and 1 mm long. These wires are
hung on the stable mounting frame of the
bridge. They will be surveyed from four con-
crete pillars erected specially for the purpose.
Control points will be placed at the front and
back of each overlap area. The survey ofthese
marks presents several problems and it will
require much careful planning to keep the
horizontal and vertical scale factors equal.
The nature ofthe box girder is such that the
direction of deformation of any point can be
fairly well predicted. This gives rise to the
possibility of using false parallax techniques
on pairs of photographs taken from the same
point before and after deformation. Since
there will be approximately 30 camera sta-
tions, it will not be possible to relocate the
camera exactly. A fairly simple adaptation of
the false parallax principle is being investi-
gated which requires that the photographs
need only be taken from approximately the
same point and with roughly the same orienta-
tion."
Rock MECHANICS
Collaboration over an eight year period be-
tween the Rock Mechanics Group at Imperial
College of Science and Technology and the
Department of Photogrammetry and Survey-
ing, University College London, led to sev-
eral applications of photogrammetry. Wick-
ens and Barton (1971) have explained how
close-range photogrammetry was used to de-
termine roughness profiles of model joints
(Figures 14 and 15). Similar model material
was then used in the construction of two-
dimensional open pit excavations. Stages in
these excavations and associated displace-
ments in the structure were analysed by false
Fic. 14. One photograph of a stereopair, taken
in order to determine surface roughness of model
joint material.
