(IBM PCs and compatibles) and the 'Macintosh world', respectively. The shift from
mainframes and minicomputers to inexpensive PCs for image processing in remote
sensing was a natural consequence of the users' desire to have a dedicated system for
personal use.
Image processing in remote sensing, however, is constrained by a number of processing
requirements, including: (1) image sizes are much larger than standard CCD images
used in robot vision and automated manufacturing applications and can exceed hundreds
of megabytes for satellite data; (2) images are usually multispectral and/or have to be
registered with other images; and (3) remote sensing images are inherently 'fuzzy' and
image interpretation based on a priori information such as solid geometric models often
used in computer vision is of limited value (Muller, 1988; Sanz, 1988). Consequently,
image processing systems for remote sensing have to be capable of large data file
handling, must make use of color display techniques, and have to enable easy and
efficient user interaction.
Full functional image processing systems for PCs usually require the acquisition of an
additional high resolution graphics board and display (e.g., 1024 x 1024 pixels x 32
bits) (Sabins, 1986; Richards, 1986). However, progresses in standard color display
PC technology starting with the introduction of the enhanced graphic adapter (EGA) in
the early 1980s has made digital image processing of remotely sensed data possible with
off-the-shelf products (Myers and Bernstein, 1985; Welch, 1989). It must be noted that
more or less all of these developments were based on IBM compatible computers since
the Macintosh series of PCs until recently abstained from color. Consequently, image
processing software for Macintosh PCs could mainly be associated with desktop
publishing or medical/biological application. With the introduction of the 8-bit graphics
option for the Macintosh-II, however, the monochrome impediment was removed
facilitating for the first time multispectral image processing on the MAC-II (Ehlers,
1990).
BACKGROUND
The Department of Surveying Engineering operates successfully a number of
Macintoshes in its curriculum. Students seem to get easily familiar with the Macintosh
interface and menu structure and are immediately fascinated by the What you see is
what you get' display philosophy. It were actually the students that suggested the use
of Macintoshes for a graduate level course in 'Image Processing in Remote Sensing' to
be taught in Spring 1989.
At the start of the course, no commercial image processing system for the analysis of
remotely sensed data was available for the Macintosh. If we were to use the Macintosh
computers for programming assignments, we had to design and develop our own image
processing software. Instead of using existing software or adding specific modules to
commercial packages, we decided to gain experiences in image processing by designing
a prototype system on the MAC-II. These practical assignments would accompany the
lectures and allow a first hand experience in putting into reality classroom concepts and
theories (Ehlers, 1990).
