International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
nurture and delivery of the information — capture,
maintenance, validation, analysis, query, elaboration,
rendering and delivery. The recent initiative (Intergraph et
al, 2004) to ensure client-neutral interoperability with Oracle
Spatial is currently restricted to 2D data but may well be
extended to 3D so as to support photogrammetric clients.
Another current trend, set to be realised in 2004, is for
photogrammetric capabilities to be made available in the
guise of familiar GIS editing tools operating in a 3D/Stereo
environment. Examples that are anticipated include Z/l
Imaging photogrammetry used via the Intergraph Geomedia
interface and BAe Systems SOCET Set photogrammetry
used with ESRI's ArcGIS. It will be interesting to see which
style of user interface proves the more popular.
These levels of integration and interoperability presently
depend on active collaboration between the different
technology suppliers. Particularly for 3D data, the broader
vision of interoperability, with photogrammetric capabilities
deployable in a Web Services architecture and ‘plug and
play’ integration are still some considerable distance away in
the future. Some of the standards issues involved are
discussed in section 5 below.
4. SOME IMPORTANT USE CASES
4.1 Update (or Revision).
Market requirements for better currency of framework data
are. increasing, especially from new business areas like
mobile navigation, urban planning and business applications.
There is in consequence more emphasis on update and or
flexible patterns of update. Updating (used here as a
synonym for revision) is the task of comparing the present
state of the database with a more recently generated source or
dataset, detecting and capturing changes and reflecting these
in the database. By means of updating the database is
regularly adapted to reflect changes in the real world. As
such, it increasingly represents the *bread-and-butter' tasking
of photogrammetric workstations. Some industry observers
(Keating, T., private communication) have gone so far as to
state that ‘The need to populate and maintain GIS databases
has driven a re-growth in the photogrammetry community’.
The trend towards richer data models in these GIS databases
is a powerful motivation towards closer integration of all
update tasks (including photogrammetry) with the database.
The richer the model, the more checking is needed, and the
greater the cost of remedying undetected errors, particularly
as contamination of the data can spread enormously. Indeed
such contamination may not be recoverable except by very
expensive manual intervention and re-doing the whole
process. Automation of the checking processes is both
necessary and to a large extent achievable, but there is a big
incentive to centralise these processes at the database level.
Two particular cases are examined in sections 4.3 and 4.4
below.
Other important aspects of Update include the maintenance
of metadata, to reflect the current status of update, and, in
many instances, the preservation of the historv of previous
states of the data. Both of these aspects are suitably handled
at the database level.
762
4.2 Refinement and Positional Accuracy Improvement.
Refinement is the process of increasing the quality or content
of existing data in terms of its geometric accuracy, its
topological structure or its thematic content (by the addition
of further attribution).
The advent of widely available high accuracy GPS
positioning has highlighted inaccuracies in absolute
coordinate positioning in many core or framework
topographic datasets. These have historically maintained
high levels of relative positional accuracy — the much lower
level of absolute accuracy having been less material. With
the wider use of such data in association with contemporary
GPS systems, this particular ‘nettle’ is now having to be
grasped by national mapping agencies and their customers.
The situation is much more widespread than might be
generally known (EuroSDR, 2004) and rectifying it will be
the cause of considerable investment over the next few years.
Suitably controlled imagery (either used in photogrammetric
workstations, esp. when 3D data is involved, or as
orthoimagery if 2D data is involved) is a primary source for
positional accuracy improvement (PAI). Almost inevitably,
in PAI programmes, changes in the data due to real world
change (update) and changes due to accuracy improvement
are generated together. It is important for users of the data to
be able to distinguish between these, in so far as this is
possible. It is easy for the data supplier to be unmindful of
the problems posed for users in modifying the positional
content of their own data, consequential on the refinement of
the accuracy of the topographic or framework data.
Unfortunately the remaining errors in absolute position are
typically unsystematic since systematic errors will have been
dealt with already — see Fig. 5 for example.
s r
RR: 1
Fig. 5. Non-uniform differences in spatial errors in US
Census TIGER data.
The provision of sufficient information (eg fields of shift
vectors) to allow users in conjunction with their software
systems to adjust their own data is a key aspect of the task,
and poses an additional requirement on any systems
(including photogrammetric workstations) used for PAI to
record and generate such information.
The PAI task for topographic or framework data may well be
of such a magnitude as to call for some degree of automation,
although to date it has been addressed in a more or less
piecemeal short-term manner. Given the present state of the
art in feature extraction, it would seem that the constrained
problem - ‘There is, very likely, a feature with this geometry
somewhere near this location in the image. Is it still there? If
so, what are its coordinates? If not, refer to operator’
Interne
should
timesc
43 S
A maj
adopte
supplk
*objeci
synon
model
increim
comm
S57st
(IHO).
data, €
Britair
nation:
datase
regulal
Increm
one of
Identif
data c
relied
appro
be def
mergir
themat
to cov
is retai
All up
enforc
respec
ones, €
mecha
In all
streng!
databa
itis
identif
proble
44 S
The d
proces
of the
repeate
(Edwa
becom
Oracle
Laser-
server-
ESRI’
the cli
which
The es
techno
establi
exploit
slivers
