International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
4.3 Avoidance of Occlusions
Dissimilarities between overlapping images are most distinct where
vertical terrain objects occlude adjacent objects. One of the
attractions of laser scanner data is their ability to cover areas
otherwise hidden in stereo models. Sufficient image redundancy can
achieve a comparable avoidance of occlusions.
5. THE *KILLER ADVANTAGE" -- ECONOMY
The cost of a digital camera operation is defined by the annual
depreciation of the camera, the survey plane, crew and flight
environment. lt is not, however, defined in any way by the number
of mages being created per year. When compared to film imaging,
there exist only the fixed costs of the digital system, no additional
costs per image made (with the exception of additional flying time if
60% side-laps are used). The more images are being made, the less
are the costs per image. Plane, crew and flight support systems are
considered the same for film and digital cameras, except for extra
fuel and extra use of the survey plane if side-laps get increased from
20% to 60%. What differs is the cost for consumables (film, photo
processing), certain labor such as scanning and film management,
maintaining a photo lab and a film archive.
Figure 3 attempts to compare the economy of a digital versus a film-
based flying operation. Of course the numbers are rough estimates
and will vary for specific situations. The cost of flying 60% side-
laps may vary greatly dependent on the cost structure of a flying
operation. The costs of an archival and cataloguing system could be
charged to the photogrammetric processing rather than the flying
operation. Generally, however, one will quickly see that digital
camera operations reduce costs, yet produce a higher value input
data set.
Figure 4 compares the costs of photogrammetric processing into
map/GIS data products such as DEMs, orthophotos and 3D vector
data. Here the numbers aim at comparing “apples with apples”, thus
the same ground areas and the same data products, once produced
by a current organization using both digital data from scanned film
plus film data on analytical/analog plotters, and in the other case a
radically modern organization with a fully digital work flow. Figure
4 assumes that all of the annually produced 20,000 film images get
converted to orthophotos, and also get converted to 3D vector
maps/GIS-input. With an annual cost for a human operator at $
40,000 the manual labor is the overwhelming cost factor, and
depreciation for equipment use is but a minor expense. The
8099/60960 overlaps in the digital domain will produce an
overwhelming advantage in automation, and the dual
analog/softcopy work flow can be abandoned, leading to significant
cost reductions.
The conclusions from Figures 3 and 4 are that there should be a cost
reduction by a factor of roughly 2 for both the acquisition as well as
processing of digital images into deliverable data products.
6. MARKET CHANGES?
Indications are that the following may happen:
e The split into image acquisition specialists and
photogrammetric processing specialists may break down.
Processing collected digital images into deliverable images
(the post-processing need) makes it logical that level 4+
processing gets merged in with the production of color images
by post processing. Also automation will make it illogical to
separate image acquisition from image analysis.
e The end users may want to consume part of the cost savings
and are going to pay less per square kilometer and data
product. Since mapping budgets will not shrink, there will be
more images and more frequent re-mapping,.
e The concern for quality may be replaced by the concern for
rapid response and lower costs, leading to a reduced concern
for quality.
e New data products will emerge that better model the 3-
dimensionality of the human habitat, and that take full
advantage of emerging mixed reality, immersive human-
computer interaction and wearable computing devices.
DIGITAL CAMERA PHOTOGRAMMETRIC ECONOMY
Item US$
Level 4+ processing depreciation Incl. in archiving
Depreciation for 6 manual editing work stations 6,000
Manual editing, 1 hr/ image! 500,000
Depreciation for 13 stereo work stations, 2 shifts 6,000
Stereo plotting, 2 hr/image! 1,000,000
SUM 1.500,000
Per each of 20,000 film-equivalent images 75
FILM CAMERA PHOTOGRAMMETRIC ECONOMY
Item US$
Depreciation of 7 analytical plotters, 3 shifts 70,000
Depreciation of 6 softcopy work stations 6,000
Manual editing, 2 hr/image" 1,000,000
Stereo plotting, 4 hr/image" ^ 2,000,000
SUM 3,000000
Per each of 20,000 film images 150
Figure 4: Comparing the costs of converting 20,000 film images and the
equivalent number of digital images into photogrammetric products. lt is
assumed that all 20,000 images get processed into DEMs, ortho photos and
3D vectors. Note that there will be 60,000 digital images covering the area of
20,000 film images. The 60,000 digital images are input into the automated
procedures. For the manual work on the digital images, only the “best” stereo
overlaps are being used. The manual work is proportional to surface area, not
number of images, and therefore is computed on the basis of the film images,
to compare “apples with apples”. Savings at $ 75/image.
In Figure 3 on film cameras, no value was attached to the
depreciation of film cameras. However, if no new film cameras get
sold, then the value of film cameras goes to zero, and any current
book value would need to get written off. Similarly, any residual
Assuming an annual labor cost of $ 40,000, and 1600 labor hours per year,
addressing 20,000 “entities”.
Manual work for AT, DEM preparation, orthophoto creation and vector
collection.
762
In
bo
ha
Ge
WC
CO!
be
prc
of
ma
RE
Lel
Dis
Ge
Gn
Ph
Ph
28:
Pet
is]
