Digital
Orthophotos For Smaller Projects? Yes You Can!
By Robert A. Fowler
With many large
institutional GIS users looking to maintain their base
data through the application of digital orthophoto
production, many smaller users are probably wondering if
this technology is really suitable for them. The short
answer is yes, it is.
The small municipality, the
civil engineering design group, the transportation
engineer and the planner can all use digital orthophoto
products as effectively as a base map for whatever purpose
they would normally use a vector map.
The primary consideration
for the prospective user will be whether their current GIS
or CAD system can accept raster data and, if not, what
they will need to update their systems. While most GIS
systems and many CAD systems will accept raster
information, there are sometimes limits. The universally
found AutoCAD, for example, requires an add on package
such as CAD Overlay GS to enable it to read a gray scale
raster. Intergraph requires the IRAS C module to be able
to use gray scale data. I made a distinctive reference
there to gray scale because there are less expensive
raster packages which will accept "document
raster" data, which are data where the pixels are
strictly black and white (i.e. the pixel is either off or
on with no gray value). To use a digital orthophoto you
need a package that will provide you with 256 levels of
data from black through various shades of gray to white.
ARC/INFO versions 6 and greater will handle raster
information and Tydac SPANS being a raster system also has
no problem with gray levels. While I have mentioned a few
of the more popular systems, many other GIS packages now
have raster capabilities and readers should consult their
manual or phone their supplier if they are not sure.
Raster files are usually
relatively simple files. A raster header usually indicates
the size of the file in pixels which is determined by the
number of pixels across (columns) and the number of pixels
down (rows). In other words a raster file could be 5,000
columns by 5,000 rows or any other combination of figures.
The number of rows does not have to be the same as the
number of columns, but the file will have to form a
rectangle. Where a chunk of photo (gray scale) data may be
missing from the file, say for instance a corner of the
photograph has been cut out, the raster image file must
still be a complete rectangle, and the missing photo data
will be replaced by an equivalent number of pixels - they
will just either be all black or all white.
When loading the data file,
the CAD system or GIS has to know several other important
pieces of information. First it has to know where the
raster information starts. This is where the first pixel
in the first column and row is located. Again, usually it
is either the top left of the file (or if you want to
think in visual terms, the top left of your screen) or the
bottom left. Then it has to know the geographic value of
this insertion point (easting and northing of the first
pixel), then the direction the raster will go from there
and the size of each pixel. For example, does the pixel
represent 5 feet by 5 feet of the ground or 1 foot by 1
foot, etc. (Pixels, of course, should be square.) Then the
rest of the raster file is usually the gray scale record
of each succeeding pixel. If any of the header information
is incorrect, say the length of the row is wrongly
entered, then the result will be the raster reaches the
point where it was told to end the line, and immediately
wraps around onto the next row, creating a garble of
unintelligble pixels.
So, much of the important
bits of information are in the header and, preferably,
should also be on the label of the tape cassette or CDROM
in case you physically have to type in the file format
information when setting up your work space in the system.
In the case of color, of
course, there are three records for each pixel: they being
the gray scale value of the red, the green and the blue
primary projection colors. As you can guess from this,
however, color files are three times the size of black and
white files. When dealing with large areas this can pose a
significant problem for data storage and file
manipulation. If it takes your system one minute to load a
black and white raster image, it will take three minutes
to load the same image in color. The upside is, in most
systems once the image is called to the screen,
regeneration of the image at different looks (zooms or
pans) is relatively quick.
While it is possible that
many companies may not have all of the pieces that they
need to be able to look at and use digital orthophoto
files, in most cases the upgrades necessary do not cost
very much compared to the overall savings which can accrue
through the use of digital orthophotos. For example,
digital orthophoto production is usually (depending on the
circumstances) less expensive than vector mapping. If a
digital orthophoto project is to cost $8,000, compared to
the cost of a vector line map at $11,000, then it is
possible to pay for the add-on software required for
AutoCAD, for instance, on one project.
The advantage of the image
file is the technical staff or engineer can see what is
really there on the ground, not what someone (even a well
meaning and professional someone) has interpreted or
simply left off the map during the mapping process. An
example of this could be flower beds or other vegetation
cover, which are not normally shown on a line map. While a
flower bed is not a major impediment to a construction
project, people have a tendency to get upset if someone
appears one day with a backhoe and without so much as a
by-your-leave destroys the zinnias planted in front of
their property!
Indeed, one of the great
advantages of orthophoto products is their use during
public meetings. A plan showing the proposed changes to an
intersection is far more understandable if the planned
road limits are dropped out (white) on an ortho
photograph. Humans are very visual animals and in short
order can recognize objects on an aerial view and are much
more easily able to appreciate proposed changes shown on
actual images rather than lines on a map which tend to be
confusing to the average citizen.
Intera recently worked with
the engineering firm Proctor & Redfern Ltd., in the
Kingston, Ontario area, on a project which involved an
infrastructure study of the Canadian Forces Base (CFB)
just outside the city. The purpose of the study was to
evaluate the state of services in the married quarters
area of the Forces Base before transferring responsibility
of providing these services to the local municipality, the
township of Pittsburgh.
Intera flew large scale
(1"=250') aerial photography, and provided assistance
on the ground control. Proctor & Redfern did all of
the field measurements and picked up a large amount of
infrastructure detail such as manholes, catch basins,
street lighting fixtures etc., using a total station
system. Precision surveying techniques were used and
coordinates were adjusted over the network with expected
accuracies to half an inch.
In the meantime Intera
completed aerial triangulation, produced a digital terrain
model and, using these with the camera calibration report,
rectified the imagery using their International Imaging
Systems' PRI2SM digital orthophoto system. The output
pixel size was 0.1 meter (4 inches) and all data were
recorded on CD ROM for use with Proctor & Redfern's
AutoCAD system with CAD Overlay GS.
Proctor & Redfern
downloaded the total station coordinates to the system as
precise positions of the detail collected in the field and
these data were laid over the image in the CAD system.
Following this the infrastructure routes were constructed
and placed in appropriate layers. Layers were provided for
electrical, storm water drainage, sewage, and water lines.
A plot file was made of various plans (water, sewer,
drainage, electrical etc.) and Intera, using an IRIS
plotter, made hard copy clear film mylars of the image
with overlaid vectors.
There were some strange
anomalies which became apparent at random spots. A number
of manholes appeared to be misplaced. The pixels on the
image were displaced as much as two pixels from the field
coordinates. Various checks of the processes (field and
digital orthophoto production) uncovered no major errors,
although from interpretation of the photography it was
obvious the area had significant changes in relief and a
couple of the manholes appeared higher than the
surrounding ground. Finally a field check was made by
Intera to verify the image data on the ground.
In every case the problem
involved a major shift in elevation. For example, two
manholes, as expected, were actually elevated: sitting on
concrete pipes about 1.5 feet above the surrounding
ground. As they were, in both cases, near the extremities
of the area of a photo used in the rectification process,
the lean of the object created displacements almost equal
to their height. In three other instances, the manholes
were on the sides of steep slopes (1:4) which were not
entirely reflected in the digital terrain model.
As a result of the
precision field coordinate survey, which created an
incredibly accurate field quality check on the digital
image, it was determined that the majority of the digital
ortho was well within the required specifications.
However, one of the things learned was that where high
accuracy results are required, a very closely spaced
digital terrain model is necessary. In addition, objects
with sheer vertical steps (with current technology, at
least) will always have some displacement unless they are
close to the principal point of the photograph.
The net result of the
project, however, was that all parties were extremely
pleased with the information and deliverables. Intera was
pleased to have proven the accuracy of their digital
orthophotos using real world ground coordinates for
precision, quality control. Proctor & Redfern proved
the technology to themselves and its application to
engineering programs. They appreciated its abundant
information in assisting their planning and product
output. The ultimate client was pleased to have easily
readable image files and a database for a significant area
in its jurisdiction. The Department of National Defense
was also pleased to have an image record of part of CFB
Kingston.
About the Author:
Robert Fowler, O.L.S., C.S.T., C.E.T., has
written on surveying and mapping topics before for Earth
Observation Magazine, as well as for a number of other
publications around the world. He is manager of marketing
for Intera Information Technologies based in their Ottawa
office. He may be reached at 613-226-5442. Perry Petersen,
P. Eng., works for Proctor & Redfern Limited in their
Kingston, Ontario office and also teaches courses in
surveying and engineering. He may be reached at
613-542-3434.
Back
|