gates, or of facets of individual pieces of
aggregate, what we are endeavouring to mea-
sure is often determined by the grainsize and
arrangements of the mineral crystals of which
the rock (from which the aggregate is produ-
ced by crushing) is composed. Facets of most
pieces of road aggregate deviate from plane
surfaces by amounts which are small compared
with the gauge of the aggregate. It is the
patterns and magnitudes of these small devia-
tions which we must measure.
Measurement of such patterns and magnitudes
may not be impossible by contact metrological
methods, but it would certainly be tedious,
time-consuming and costly.
In studying the macroscopic textures of pave-
ments, while units of portrayal change by a
factor of a thousand - from micrometres to
millimetres - comparable comments apply. That
is, road surfaces deviate from plane surfaces
by amounts which are small in relation, for
instance, to magnitudes such as the foot or
the metre. And we remain interested in por-
traying the patterns and magnitudes of these
small deviations, as complex as those of
aggregate facets. And again, measurements of
these patterns and magnitudes by contact met-
rological methods would be tedious, time-
consuming and costly.
Non-contact Metrology Means Remote Sensing
In contemplating non-contact metrological
methods, there are not many choices. Optical
methods are available, of course, using micro-
Scopes for which movements of the optical
axis can be measured and recorded. Or it is
possible to use a narrow light-slit and to
photograph its trace on a surface. Or it is
possible to acquire imagery of the surfaces
to be studied and to take or to perform
measurements using such imagery.
Photographic imagery is that most familiar,
in which use is made of a lens System and of
an emulsion which is sensitive to some seg-
ment or segments of the electromagnetic
Spectrum. But there are other types of sen-
sors using linescan techniques, as have been
developed for use in satellite and airborne
studies of characteristics of the earth's
surface or the surfaces of the moon and the
planets. Another linescan sensor is the
scanning-electron-microscope (SEM) which has
great versatility in producing imagery of
surfaces at magnified ratios up to many thou-
sands. Whether we use photographic sensors or
Scanning sensors, the methods come under the
general category of remote sensing.
The Photogrammetric Method
The geometry of the remotely-sensed image is
important in its measurement. From a single
image, whether it is compiled by a series of
linescan images or by the instantaneous ex-
posure of a photographic image, we can gene-
rally establish one spatial direction between
a point in space (which can be identified) and
any image point. To do so we must know the
geometry of the "bundle of rays" which con-
stituted the image and either reconstitute
that bundle or understand the consequences
of failure to do so.
In topographic photogrammetric practice, the
Systems used include aerial survey cameras of
102
known calibration and restitution instruments
which can reconstitute the bundles or rays of
constituent photographs of stereo-pairs within
known tolerances. With two images of any point,
two spatial directions can be established and
the intersection of those two directions de-
fines a unique point in space.
With two overlapping images (a pair of con-
jugate photographs) the procedures of "rela-
tive orientation" involve a spatial resection
with five degrees of freedom. Following rela-
tive orientation there is in the photogramme-
tric instrument a spatial model of the object
or scene photographed, which closely approa-
ches the spatial geometry of that object or
Scene at some arbitrary scale and with an un-
known spatial orientation. For the purposes
of studying surface textures, interest may be
restricted to establishing the absolute para-
meters of scale. (In topographic photogram-
metry it is usual to establish the other ab-
solute parameters of azimuth and gradients in
two axis directions, plus the coordinates of
one point, but those six parameters are of no
interest for the purposes introduced in this
paper.)
Where use is made of image acquisition sys-
tems which are not geometrically calibrated
cameras, the whole system of image acquisi-
tion and evaluation needs to be calibrated
for an appreciation of the significance of
observations taken with such a system. In the
work reported in this paper, no attempt has
been made to calibrate the image acquisition
Systems separately, but an assessment of the
significance of observations taken by the
whole image acquisition and evaluation sys-
tems has been attempted. Relevant procedures
will be described in some detail.
An aspect of these procedures which deserves
emphasis is that, with no exact knowledge of
such parameters of the image acquisition sys-
tem as its principal distance (focal length)
it must be assumed that the spatial model
produced in the photogrammetric restitution
instrument is an affine model. That is, the
scale in planimetry must be assumed to differ
from the scale in height.
IMAGE ACQUISITION SYSTEMS
Angular Fields in Parallel-Axis Stereograms
An objective in initiating and pursuing the
studies of textures undertaken was to avoid,
at least in the early stages, the purchase of
expensive items of equipment. It was aimed to
use either inexpensive items or items to which
ready access was available. A problem which
soon became evident was that spatial measure-
ments in photogrammetric instruments rely on
stereoscopic parallax and that the angular
fields of camera-microscopes are very narrow.
Thus, although facilities existed for moving
the objects photographed by the camera-
microscope system, to give any desired over-
lap (such as the sixty percent which is nor-
mal in aerial photogrammetric practice) doing
So did not result in significant stereoscopic
parallax.
As was mentioned in the acknowledgements, this
work was pursued in collaboration with the
personnel of the National Measurement Labora-
tory of the C.S.I.R.O in Sydney. Arrangements
were made in that Organisation for manufac-