AUTOMATED WATERFOWL CENSUS 
Dr. D. Bajzak 
Associate Professor of Engineering 
Memorial University of Newfoundland 
St. John’s, Newfoundland 
INTRODUCTION 
The most Important uses of photo interpretation in 
wildlife management are the censusing of animal popula 
tions and the mapping and evaluation of wildlife habitat. 
In some directions satisfactory photographic techniques 
have been deveoped, in others much remains to be done. 
Among the latter is the need to develop techniques in 
photographing large number of waterfowl and a rapid and 
reasonably accurate method for counting large numbers of 
individuals on the photographs taken. 
From 1967 serious attempts were made by the writer 
and personnel of the Eastern Region of the Canadian Wild 
life Service to develop aerial photographic techniques 
for the census of game birds. Large flocks of birds were 
photographed through the porthole of an aircraft. Then 
these photos were used to count the birds, if possible 
by classes. Visual observation of some photographic speci 
mens indicated that at least three, but possible four classes 
of Eider ducks, Somateria mollissima , can be distinguished. 
This suggested that with the use of a scanning microdensito 
meter the class distinction would be more objective. In 
addition to this, the counting procedure could be automated. 
While the development of the aerial photographic proce 
dure is relatively easy, the development of an automatic 
counting procedure is very difficult and time consuming task. 
The bird images first have to be digitized, then valid patterns 
have to be established, finally bird images have to be separated 
by comparing them to pre-established patterns. 
In 1971 research work was initiated by the author in 
the field of automated waterfowl census using aerial photo 
graphy and microdensitometry. From 1971 to 197^ this research 
was supported by the Canadian National Research Council which 
is acknowledged.