Many vendors offer products with limited functionality, for
example the popular orthophoto engines which do little more
than generate orthophotos from the user's raw images and
DTMs. Included here are five of the vendors mentioned above
- Galileo SISCAM, ISM, KLT, Kork (now Vision) and Matra -
but more recently the remote sensing software suppliers have
ventured into this area, for example PCI have DTM and
orthophoto modules, both for aerial photos and for SPOT
imagery, forming part of its EASI/PACE package, and Earth
Resource Mapping are extending the photogrammetric
components in their ER Mapper suite. TNT-MIPS from
Microlmages is another well known remote sensing and image
processing package that offers orthophoto and DTM
capabilities and will be extended to stereoscopic viewing.
Lohmann ef al. (1990) and Dowman (1991b) attempted to
grapple with this rich variety; the latter proposed a simple
classification into four categories:
A performance and function of an analytical plotter
with automatic feature extraction
B performance and function of an analytical plotter
with image processing and computer assisted feature
extraction
C systems designed for specific applications with high
performance but limited functionality
D limited performance and functionality but low cost.
Today we still value such an approach, but perhaps a very
practical taxonomy has merit too. Note that the above
paragraphs represent a most rapid market survey of systems
known to the authors. Doubtless some important systems have
been omitted. We do not survey systems in academic and
research institutions which are not widely offered on the open
market and little attention is paid to systems aimed purely at
close range photogrammetry. Five categories appear to emerge:
1 DPWS from major players offering the great majority
of photogrammetric functions: Intergraph, Leica-
Helava, Matra, Vision International, VirtuoZo, Zeiss
2 DPWs from smaller vendors offering less than
complete functionality, but with great strengths in
limited areas, such as feature extraction; often these
workstations were designed primarily to emulate
APs: DAT/EM, ISM, KLT
3 systems tuned to one function, which is offered at an
economical price; these systems do not aspire to be
fully functional DPWs: Vexcel Imaging
(triangulation), Vexcel (close range), ISM
(orthophoto), Vision International (orthophoto)
4 photogrammetric modules from the remote sensing
vendors, usually offering orthophotos and perspective
scenes and, less often, DTMs: PCI, ERDAS, Earth
Resource Mapping, MicroImages
5 systems with limited functionality, yet still offering a
wide selection from the DPW spectrum of
operations, all at a low price (DVP, R-WEL, VTA).
1.4 The last four years
As part of the historical process outlined above, the vendors
mounted an aggressive display at the ISPRS Congress in
Washington, D.C., in 1992. The photogrammetric community
awaited the rapid adoption of the new technology, yet the
Congress was followed by a long period of slow sales. Perhaps
this was a time of contemplation when users made up their
minds whether to switch or to continue with AP technology
(Petrie, 1992); perhaps it was a cyclical dip in the G7
economies. In due course there was a recovery and buyers
became active again but, intriguingly, both analytical and
digital systems were sold in greater numbers as a result. In the
386
period 1994-96, Leica SD2000/3000, Zeiss P3/33 and Adam
Promap APs sold in large numbers, enjoying a trend that if
anything was slightly upwards, but simultaneously many users
turned to digital and there were even a few who jumped
directly from analogue to digital One factor was the
traditional suppliers" natural protectiveness towards their core
technologies, resulting in aggressive selling and highly
competitive pricing. Another was the dominance of Intergraph
in the early 1990s and the wish of its competitors to close the
gap. And a third was the realisation on the part of users that
digital photogrammetry enabled them to offer a greater range
of deliverables to their clients, who in turn were ready for
products like DTMs and image maps and were often in the GIS
business and happy to populate their raster layers.
2. HARDWARE
Section 2 describes the hardware components of a DPW,
whereas discussion of the software is held over until section 3.
2.1 Principles and basic components
A DPW consists of a graphics workstation with enhanced
image processing, memory and display capabilities including,
in most but not all cases, a facility for stereoscopic viewing.
2.2 Host computers
A powerful processor (CPU) and a very large memory are
required to handle the large volumes of image data. One can
discern trends through the four years, for example in the
computer platforms used and in the range of functionality in
the software. Photogrammetrists have been complaining since
the 1950s that the available computers are insufficiently
powerful for their needs, but both workstation and PC
platforms have developed to the extent that they are now
adequate for most aspects of digital photogrammetry. Raw
processing power, for example, allows DTMs of half a million
points to be generated in under an hour, or orthophotos to be
produced in a few seconds per megabyte. The most popular
hosts are the Intergraph Clipper, Sun SPARCstation and Ultra,
Silicon Graphics (SGI) Indy and Indigo? and top end PCs,
usually Pentium-based. Many earlier machines, such as
SPARCstation 2s and 10s and 80486 PCs remain in use.
Typically 64 MB of RAM are recommended, though in some
cases the methods used to handle the graphics for stereoscopic
viewing make greater demands (see 2.3 below). The majority
of these hosts operate under some brand of the Unix operating
system, such as Intergraph’s Clix, Sun’s Solaris or SGI’s IRIX.
The PC solutions run mainly under Windows 3, but at least
one vendor prefers DOS and some of these systems are at the
time of writing being ported to Windows NT or 95.
2.2.1 Accelerators. In the past it was customary for DPWs to
contain extensive custom built electronics The DSCC
mentioned above had very few off the shelf components and
even its successor the DF/s (Digital Extraction Segment), of
which DMA took delivery of several hundred from 1986 to
1993, relied on custom built disk arrays for data storage. There
have been units which have used digital signal processing
chips (DSPs) or array processors to cope with the
computationally intensive processes required for image
matching, but the latest workstations have enough power to
handle these with off the shelf components.
Much more common, however, are graphics accelerators. A
weak point in DPWs has always been the special hardware
necessary for image panning and stereoscopic display.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
AY IN s £s ^ AM m LY
