verted for x, y digitization by a rack-and-pinion
rotary optical encoder, an electronic data-dis-
play console, and a key punch. A software pac-
kage called TAPS (Texas Automated Plotting
System) was designed for the following func-
tions :
1. Editing and special routines : In addition
to being edited, the machine x, y data are
transformed to the ground system.
2. Storage : Constructs file from edited data
and stores on disk.
3. Processing: Either total or selective gra-
phics for land-use information and classifica-
tion levels ; variable-scale format for route ana-
lyses; perimeter and area computation; and
environmental resource data for corridor access
studies.
A key-punch collection system limits rapid
editing and reduces the efficiency of the system.
Therefore, the key-punch collectors are to be
replaced with multiple automated stations.
Each station will consist of 3-5 stereoplotters
with a CRT display and a minicomputer. The
multiple automated stations will be collected
into a comprehensive automated information-
management system.
Terrain data file for route location
The Texas Highway Department has also
combined a numerical ground-image system
with a roadway-design system (Crawford and
Guess, 1974) by developing a scan-profile nu-
merical terrain-storage system and adding to
the road-design system a cross-section request
and retrieval system. The scan-profile system
produces a numerical representation of the ter-
rain surface from parallel-scan profiles obtained
by conventional photogrammetric methods.
Once a data file has been built for the terrain
surface, it can be used to generate cross sec-
tions along any specified base line that has been
previously stored. This system provides the de-
sign engineer with a means of studying designs
of alternative locations without cross sectioning
each proposed alinement or repeatedly cross
sectioning alinements that are changed.
Automated rectification
The Utah Agricultural Stabilization and
Conservation Service has developed an automa-
ted system for producing rectified scaled pho-
tographs (Dickson, 1974). Hardware consists of
18
a point-transfer device, an H. Dell Foster mono-
comparator with digitizer and paper tape, aad
a NOVA 1200 minicomputer. Software routi-
nes are image refinement, three-photo relative
orientation, polynomial block adjustment, sin-
gle-photo resection, and computation of rectifi-
cation parameters. An H. Dell Foster autorecti-
fier uses the parameters directly to produce the
required prints. Working scales of the obtai-
ned photography (from 6-, 8 1/4, and 12- inch
cameras) range from 1:20,000 to 1:40,000.
Universities and Private Organizations
Digital image processing and analysis
A number of excellent articles on digital ima-
ge processing have been presented by represen-
tatives of private American companies. In ge-
neral, these refer to three principal phases: data
collection (A/D conversion when source is ana-
log), digital processing (D/A conversion when
output is analog), and output (Helava et ai,
1972). The processing phase may include :
1. Preprocessing, such as noise filtering, en-
hancement, and coordinate transformation.
2. Shaping or rectifying to change geome-
tric shape and remove deformation.
3. Image matching to identify corresponding
imagery.
4. Pattern recognition to classify data.
The output concerns either dimensional
characteristics, as in a digital terrain model, or
pictorial characteristics, as in an orthophoto-
graph.
A study by the Control Data Corporation
(Lillestrand and Hoyt, 1974) points out that in
designing a digital image-processing system the
following conditions should be considered :
1. Digital size of encoded analog image
2. Nature of digital image transformation
3. Requirements for display, both tonal and
alphanumeric.
The authors conclude that in accuracy clas-
sical analog techniques for measuring and gene-
rating film images are superior. On the other
hand, they regard digital techniques as supe-
rior for image processing because of the ability
to perform complex transformations. The ex-
ceptions is the Optical Fourier Transforms,
which are faster than digital transformations.
