omly
igital
from
copy
770,
1ance
were
cally
to 2
'Train
ions)
ieved
scale
DTM
ights
nages
n the
JTMs
h the
y, as
other
itute,
s of
maps.
ality,
d its
e the
fering
ration
and
on is
stable
from
is and
mated
nd in
metric
le one
iming
ve can
tegies
(e.g.
with
rends,
ith the
erator-
k ofa
nd is
-prone
rming
of an
object
uiding
and to
providing logical decisions (e.g. is this a road or a river?),
which are the most difficult to successfully automate.
The advancements in the above research topics are supported
by developments in softcopy photogrammetric
workstations [Heipke, 1995], which are slowly but steadily
gaining the trust of practitioners. Their advantages can be
summarized as
® operational ease, with cumbersome observation and
measuring tasks substituted by computer-executed
tasks,
e versatility, as limitations associated with analog and
(less so) analytical instruments are naturally bypassed
by softcopy systems, while at the same time the
compatibility with other parts of the geoinformation
environment is optimized, and
e cost-effectiveness, as they are less expensive than their
analytical counterparts, and in addition offer, through
automation, better time performance in their intended
operations.
When examining the performance of digital orientations,
aerotriangulation and other operations, and comparing them
to analytical processes, one must not ignore that, beyond
accuracies, operational ease and user-friendliness are
essential issues influencing the choice of the practitioners.
Thus, while some of the above operations (e.g. orientations)
are performing accuracy-wise equally to analytical methods,
the immense potential offered by automation (e.g. the
simultaneous measurement of hundreds of conjugate points
in a stereopair within few minutes) make digital operations
overall superior to analytical ones.
4. EFFECTS OF INTEGRATION ON GIS
From an image analysis point-of-view, and beyond the
obvious practical database issues associated with the
integration of large raster files and relevant vector/object
data in a GIS, the most important effects of photogeographic
integration on GIS can be classified under two broad
categories, namely GIS multidimensionality and the
integration of accuracy infornration.
4.1 GIS Multidimensionality
Currently, typical GISs operate on a 2.5-D mode, with a
single z-value attributed to a point (x,y), often through the
use of DIM information. Within an integrated
photogeographic information environment we are moving
to full 3-D data, and furthermore, by considering the time
parameter of data, to multi-dimensional operations. The
move towards fully 3-D GIS is also supported by the
potential for the fusion of aerial with terrain digital imagery
and 3-D building models extracted from it [Streilein, 1994].
The move to 3-D GIS results from the integration of 3-D
object information (e.g. buildings or 3-D vectors) extracted
through image analysis methods. The transition from
existing 2.5-D to full 3-D GIS is much more complex than
simply adding another layer of information, which does not
constitute integration [Fritsch, 1990]. Full 3-D database
operations, like queries and visualization processes, would
27
not be covered by such an extension. From a practical point
of view, the extension of an already functioning GIS to
accommodate a third dimension is deemed non-trivial, and
database storage and management concepts and
methodologies need to be properly modified to support this
extension. This can be extremely difficult even for versatile
object-oriented systems, thus suggesting that the
development of novel prototypes appears to be a more
appealing solution.
Within this broader concept, 3-D object representation is
important. Boundary representations (B-Rep) are very
suitable for 3-D objects, especially employing CAD, but
spatial occupancy enumeration, constructive — solid
geometry, and cell decomposition in general (or octrees in
particular), are valid alternatives for 3-D object structuring
[Fritsch & Schmidt, 1995].
In addition to the third topographic dimension, the
integration of digital imagery is emphasizing the role of
time as a fourth dimension within GIS, thus making
integrated photogeographic environments actually 4-D.
Even though the temporal aspect is inherently included in
current GIS applications, the use of imagery, which by
nature is time-specific, is making its exploitation more
pressing in database management systems (DBMS). Within
this framework, geoobjects can be described by their spatio-
temporal extent and behavior [Shibasaki, 1994].
4.2 Integration of Accuracy Information
Currently, photogrammetric data are typically treated within
a GIS as deterministic values, ignoring spatiotemporal
geometric and thematic uncertainties associated with:
e the methodologies used for their production (e.g. the
algorithm employed for the generation of a DTM or the
measurement of an outline),
e the quality of the data employed within these
methodologies (e.g. resolution and sensor
characteristics of digital imagery employed by the
above methodologies), and
e the temporal validity of these data (e.g. date of capture
of the imagery which was processed to produce the
metadata of interest).
An integrated photogeographic environment is characterized
by the multitude of data and associated sources and
algorithms. Within such an environment, the proper use of
information requires the identification of the uncertainty
estimates associated with it for proper error propagation
analysis within database operations. Thus,
photogrammetrically produced data can be viewed as a
specific form of fuzzy information within a GIS: they
express information which is not inherently fuzzy (i.e. the
outline of a building, or terrain heights), but is available
with some measures of accuracy (and consequently,
inaccuracy) associated with it. This is one of the critical
issues differentiating the integration of digital
photogrammetry vs. remote sensing within a GIS, as remote
sensing is typically dealing with inherently fuzzy entities
(e.g. outlines of cultivated areas).
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
