2, easy and
zed digital
tories as a
and sites
ammetrists,
A SMALL
2 FOR
.CORDING
=M
uested by a
tudies and
comprises
ontal cross-
and walls,
f 1: 50 or
} to::4 1.14),
periodical
ntation of
er another.
n recording
art, and in
of pictures
vhat makes
ations. The
of historic
t decrease
rding team
the hand-
hots which
the historic
5 of plotting
1e foresees
] contents.
tion should
ntation can
c structure
,
dominating
'al record.
ne fact that
but with a
Interesting
jroved the
ain type of
t likely to
resentation
This very
wing and
hotography
y demands
/ssed need
letails of a
cord has a
free choice of proportions when presenting the structure
and can show a drawn constructional line on the
background of photomosaic or, on the contrary, the
picture of a detail can be exposed on the background of
the drawing. A desire to make photointerpretation
possible may prompt the editor to the very vast use of
photomosaics. It should be considered, however, that
due to the better geometric interpretability of stereopairs
rather than single pictures, the usable linear accuracy of
photomosaics is a degree lower than that of the drawing;
on the other hand, the price of photomosaic does not
depend upon the density of details to be shown, and
therefore pictorial presentation may be for some
monuments (e.g.richly decorated) much cheaper than the
linear one.
Therefore a general rule can be adapted, that
constructional lines of a structure should be presented
rather in a vectorial manner, while decorative details can
be very well presented as a half-tone insertions. In some
cases [Jachimski 1975] it would be enough just to
supplement a fotoplan of a facade with some plotted
lines, just to better define openings or other
constructional details.
À very interesting examples of combination of rectified
photographic image and linear object representation ware
presented at the ISPRS Washington Congress [Pallaske,
1992].
Presently available powerfull interactive photogrammetric
work-stations provide for a very easy screen presentation
of various combinations of vector and raster technique of
object evaluation. That way the final selection of
configuration of each portion on an object documentation
can be verified visually, to the best benefit of the
customers.
Present photogrammetric workstations are extremely
universal and flexible - not all that flexibility is necessarily
indispensable, however, for producing proper
documentation of a historic monument. For those
applications extremely important would be reasonable
price of an efficient enough system, which would allow
for photogrammetry direct application by various
monument recording and restoration teams.
Let us try to determine the functions of photogrammetric
Workstations which are indispensable for historic
monument recording.
As for the recording of monuments variety of cameras
can be used - the devoted workstation must accept any
metric or nonmetric pictures. To simplify the task we can
assume that one of PC-instalable terratriangulation
systems is available (eg.ORIENT), and therefore the
discussed historic monuments documentation devoted
workstation (HMDW) must just accept pictures having
sufficient ground control information. This can be easily
executed on PC as two-steps or one step orientation of
stereopare furnished with interior orientation elements or
as one step DLT orientation for those non-metric
stereopares which were not initially processed within
terratriangulation net.
Our small HMDW must provide though a resolution good
enough to interpret and survey even very small details of
historic monuments. It is obvious that high resolution of
digital image (that of 2000 dpi to 4000 dpi) requires not
only a lot of disk memory, but also quite powerful screen
graphics. The on screen image visualization must be so
designed, that on medium screen format (14"-15") must
be visible large enough portion of the surveyed
261
monument to allow for proper general interpretation, and,
at the same time, the details must be presented with a
very high resolution (what requires a lot of the screen
pixels). That dilemma can be solved the same way as in
other systems by providing small portion of high
resolution image in a zoom window with the smaller
resolution image in the background. If it is steady image
based system (dynamic floating mark instead of scrolled
image), it can function with no need for very powerful
screen memory. To provide the operator with an easy
approach to the image portions which are out of the
screen, there would be enough to use a large RAM. In the
RAM memory the image portions fitting the screen size
(or part of it) would be stored (virtual screen) to provide
very fast image change and visualization of portions
close to the actually visible ones. The RAM of course can
not be big enough to accommodate full stereopare
images. Therefor such a rotary system must be adopted,
that the RAM stored image portions always surround the
portion currently visible on the screen. After several RAM
pre-stored image portions have been used, the RAM
stored images must be rearranged; that requires some
extra time, but with relatively large RAM does not have to
happen too often. Not only the main scale image
exchange on the screen relies on the PC RAM in such
system design. Also the zoom window can use the image
portions taken from the image pyramid (which could also
be ZOOM-ed on purpose).
So, in proposed solution the universal-use RAM rather
then the screen memory is used, what allows to use
typical off-shelf PC sets with hardware flexible for other
programs. The proposed HMDW uses image pyramid in
which the highest resolution image is predestined for
ZOOM-window, one of intermediate images is used as a
basic full screen visualized image, and the pyramid top
image of smallest resolution should fit a half screen size
and be used for the free selection of working stereopare
portions (those which are visualized for evaluation). It is
obvious that behind the relatively large RAM (not smaller
than 8 MG) would have to be a big hard disc available.
To easy deal with many high resolution images (one 6 x
6 cm image requires for true colour at 2000 dpi some 75
MB memory) it would be advantageous to furnish the PC
system with exchangeable hard disks of short data
transfer time rather than magneto-optical or similar
diskettes.
Altogether the above system may work perfectly, but
being based in a large extend on the PC/RAM, it does not
go well with windows operation system on the middle
size PC-computers. But lack of the common user friendly
communication system would make our HMDW difficult
to learn by all those users majority accustomed to use
windows operation system. Therefore intermediate
solution must be adopted, and an extra programmed
user friendly communication system similarly operated
as ,windows" should be introduced. Of course such a
windows-like interface would not have all the features
typical for real windows operating system but would use
less RAM. It would provide though, an easy enough
communication with the HMDW program. Easy -
because working on principles well know to the majority
of users.
The system stereovision working on principle of steady
image split screen stereo perception is probably the most
easily fitting the natural fizjological men vision system,
and - no doubt - the least expensive at present.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996
