61 
excited and worked extremely hard as they recognized that the 'by-product' of the class 
turned out to be a prototype for a 4th generation image processing system combining 
state-of-the-art interface design with powerful image processing techniques. All 
feedback from the students indicate that they not only 'learned a lot but also had a lot of 
fun' during the course. 
Classification 
Hide Training Rreas 
Saue Training Hreas... 
Load Training Areas... 
Parallelepiped 
Minimum Distance 
Mammum Likelihood 
Shorn Results 
Split Screen 
Figure 6: The Dirigo classification menu allows for saving and loading of training areas 
and provides three classification algorithms: parallelepiped, minimum distance, and 
maximum likelihood. 
We will continue to use the Macintosh and the Dirigo system as a tool in remote sensing 
and image processing courses. Currently, we are working on additional Dirigo 
components including: (1) vector overlay from GIS or DLG files; (2) automated creation 
of polygons from classified images; (3) image compression; (4) unsupervised and 
hybrid classification algorithms; (5) implementation of stereocorrelation techniques; and 
(6) hardcopy output. 
We see applications of the system primarily in the educational area. Potential users, 
however, might also come from the private or public sectors from various fields such as 
surveying and mapping, urban and regional planning, natural resource management, and 
growth monitoring. 
ACKNOWLEDGEMENTS 
The author would like to thank J. Jackson, P. Haggerty, M. Fuller, E. Lange, D. 
Pullar, G. Surrette, D. Steiner, C. Winne, M. White, and M. Zhao for their enthusiasm 
during the development of the Dirigo system. The support of Apple Computer, Inc. for 
this project is gratefully acknowledged. The software is marketed by PCI Inc., 50 West 
Wilmot Street, Unit 1, Richmond Hill, Ontario, Canada.