- 140 -
Density Patterns
A portion of a 9" by 9 " aerial photograph of Greater
Snow Geese (Figure 1) was digitized using 100 micrometer
scanning aperture. This yielded a sample point matrix of
400 by 600 (240 000 points). The digitization required
approximately half a minute scanning time on the densito
meter .
With visual observation of the photograph three classes
of waterfowl can be distinguished without ground truths
(Figure 1):
1) Adult Snow Goose (at B)
2) Juvenile Snow Goose (at C)
3) Unidentified species (at A)
The same classes can be defined in the digitized image
field by their size and density values. Figure 2 illustrates
the digitized images of the unknown species with background
densities. The values of the background can be eliminated
showing the densities of the bird images only (Figure 3).
It can be observed from these figures that an adult Snow
Goose has photographic density values between 74 and 97
while similar values for the unidentified species are between
119 and 139- These range values were chosen somewhat arbitrarily
The size of the bird is indicated by the number of density values
occurring within the stated ranges.
Automatic Counting
The digitized image field of an aerial photograph is
read into an IBM 370-155 computer and stored on a disc. A
FORTRAN program was developed to identify the required bird
density patterns, give some statistics on the density values
of the bird image, and count the number of occurrences of a
specified pattern. The size of image field that can be
processed at a time in the computer depends on its memory
storage capacity (or its availability). For this experimen
tal work a matrix of 30 by 400 was used. This corresponds
to a 3 mm. wide and 40 mm. long scanned strip on the aerial
photograph. At present the FORTRAN program is in its devel
opment stage therefore it will not be explained in this paper.
The programming will be completed in the near future and
results will be published in due course.
Figure 1. Aerial Photograph of Water Fowls
