----------------------------Original message----------------------------
 
This message is in response to the questions and comments of Sue
Haffner and others regarding the digitization of aerial
photography.
 
After a great deal of experimentation we at the Map & Geography
Library at the Univ. of Illinois are about to begin an expanded
pilot project involving the digitization of our pre-1950 aerial
photography.
 
Our goal in digitization was (and is) to get the best
reproduction possible for a permanent archive.
 
I collaborated with a remote sensing analyst from the Illinois
Natural History Survey and a photo interp. person from the
Illinois State Geological Survey in the experimentation.  A
detailed discussion of this experimentation is too long for an
email discussion list.  But I will try to give a concise list of
our findings.
 
1.  We are extremely happy with the results at this point. Our
old prints, in general, support a scan rate of 42 microns per
pixel giving us a ground resolution of 3.8 feet per pixel.  This
compares to 3.28 ft (1 meter) for the DOQs.
 
2.  We have developed a consistent scheme to determine optimal
scan rates which can vary for different flight dates and depend
upon quality of the print, the paper, and other factors.
Negatives or transparencies can support, in general, higher scan
rates.
 
3.  The equipment of choice is a Crossfield drum scanner.  The
Crossfield starts at $250,000 and requires a great deal of
maintenance.  Thus, we decided that we would not buy one.
 
4.  The primary companies that are experienced with aerial
photography are those that contract with the USGS for DOQ
production.  We solicited samples from three companies, two very
large companies and a young local company with a Crossfield.
None of these were able to satisfy us with the quality we were
looking for as a first effort.  However, the local company,
(Scantech Color Systems) worked very closely with us to satisfy
us, investing many hours and dollars in the experimentation, and
is prepared to handle aerial photo digitization requests outside
the state as well.
 
5.  Scantech has given us an estimate of $34 per print for the
pilot project, (a cadillac scan for an escort price), and expects
the price to go down after the pilot.
 
4.  The pilot project will consist of a single county.  DOQ
production for that county will be completed in December.  Using
the Digital Terrain Model resulting from DOQ production and the
DOQ, we expect to have the old photos rectified, and are
soliciting samples and prices from all the companies that
contract with the USGS for DOQ production and two Canadian
companies as well.
 
5.  We (our remote sensing analyst) registered two photos to
their corresponding DOQ as part of the experimentation.  I'm not
sure it is possible to overstate the effect of seeing the 1939
and 1988 photos overlayed.  The two images were compared
mathematically so that change and intensity of change could be
displayed in different colors on the screen.
 
6.  We presented the results of our experimentation with numerous
slides a la digital camera at the "GIS in Illinois" conference
early this week with very positive comments.  At this point we
have four other counties who are interested in funding the scans
for their counties, and various other large scale funding
arrangements are being discussed.
 
Sorry this is so long; if anyone wants more information or a more
detailed report, please let me know.
 
Li Hunt
Map & Geography Library
Univ. of Illinois
[log in to unmask]