142
CONCLUSIONS
The results of this study indicate that remote sensing
with pattern recognition is a very useful tool in waterfowl
census. With the computer processing of digitized imagery
the counting can be speeded up considerably and the species
can be classified into subgroups. The developed FORTRAN
program can process digitized images of species having dis
tinctive uni-color (greater snow goose, black duck, etc.)
However, our research work will cover the processing of more
complex patterns (images of Canada goose, diving ducks, etc.).
The aerial photographic techniques has to be modified and
adapted to each major species. Spectrozonal photography
will be necessary for some species to eliminate background
and to obtain maximum distinguishable patterns. More accurate
identification of species will require extensive ground truthing.
LITERATURE CITED
(1) Bajzak, D., 1972 Signature Patterns of Game Birds.
Report submitted to the Eastern
Region of the Canadian Wildlife
Service. St. John’s, Newfoundland.