FRIC
ineering,
nn
n and GIS
nce and/or
ached as a
formation
multi-tier
operability
uted web-
positional
grity. The
'om 2D to
ole of the
e données
e données
n est faite
ractérisés,
té entre le
upport de
inalyse et
isation au
l'entretien
ible pour
à 3D Le
le rôle du
erfassung
bung von
wird die
chiedene
ind feste
uren. Die
tung der
Ib eines
üblichen
entlichen
: für 3D-
nach 3D
t einigen
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
1. INTRODUCTION
1.1 Rationale of Paper
The start point for this invited paper can be summarised by
the following extract from the ISPRS Annual Report 2003
(ISPRS 2003) covering the work of Intercommission WG
II/TV: Automated Geo-Spatial Data Production And Updating
From Imagery:
“Concerning updating of GIS, besides the data capture itself
the management of the updating information in the database
is a relevant topic itself. Automatic update including
topological changes still is nearly unsolved. In some cases
from the operational point of view the acquisition of the
complete data set still seems to be easier than to incorporate
acquired changes into an existing data set.............. Digital
photogrammetric workstations more and more approach a
GIS leading to integrated solutions which cover the complete
process from data capture to data management, analysis,
visualisation and dissemination. At present they are
incorporating database and visualization functionalities,
partly in 3D. In general the cooperation and exchange
between GIS and Photogrammetry still is rather low,
especially from the commercial point of view. There still is a
lack in standardized exchange between the respective
systems but the companies seem to have recognized the lack
in integrated solutions for the end-user.”
The importance of the shift in requirement, from the
acquisition of complete data sets to the incorporation of
acquired changes into an existing data set was highlighted for
the author by a remark at the OEEPE/ISPRS Workshop
"From 2D to 3D - Establishment and Maintenance of
National Core Geospatial Databases" , Hannover, Germany,
October 2001. In describing the history of the Topographic
Database of Catalonia, 1:5,000 (Pla et al, 2001) Josep Lluís
Colomer said that in 1996 they had done something ‘they
would never be allowed to do again — they had abandoned
the previous data set and started again'. This represents the
new reality for most providers of topographic data. The
investment in the current data holding, and its increasing
richness in attribution and structure as well as topographic
detail, means that there is in practice no alternative but to
update (or revise) it, to enrich it and on very rare occasions to
re-engineer it. The only exception to this trend is the case of
*mission-specific' datasets, which may be of sufficient
importance to bear the cost of one-off creation from source,
but even in this case the trend is towards intensification of
sustained and maintained ‘framework’ data.
1.2 Structure of Paper
Section 2 surveys the reasons behind the trend towards the
increasing integration of photogrammetric data acquisition
and GIS databases, describing both the drivers and the
resulting benefits. The shift from primary data acquisition to
the maintenance and/or enhancement of increasingly rich
datasets is discussed, and the assertion is made that this
should increasingly be approached as a conflation issuc.
Section 3 enumerates and characterises different levels of
system architecture, including file-based information
exchange, loose- and close-coupling (with differing balance
of functionality between client and server-side) and multi-tier
architectures. The role of the database in supporting
759
transaction management is discussed, as is the importance of
interoperability between all the maintenance, quality
assurance, analysis and delivery processes either within the
firewall or across a distributed web-based system.
Section 4 covers a number of key use cases over and above
the standard update or revision task, including positional
accuracy improvement, the maintenance of public persistent
feature identifiers and the maintenance of topological
integrity.
Section 5 discusses the level of standards support, or lack
thereof for the integration of photogrammetry and GIS
databases. The level of support available for 3D information
in geospatial databases is reviewed together with the issues
of migration from 2D to 3D.
The paper concludes with a summary and some reflections
on the future role of the practitioners — the
photogrammetrists, image analysts and geomaticians of the
future.
2. DRIVERS AND REQUIREMENTS
2.1 The Drive to Database-Centric Operations
The fundamental requirement for the database management
system (DBMS) is of course that of being a secure and
widely accessible repository for the geospatial data gathered
at such considerable cost. At the same time as the case for the
integration of photogrammetry with GIS is increasing, the
GIS vendors are making use of mainstream DBMS
technology for its security, availability, scalability, archiving,
transaction management, query support and enterprise-wide
characteristics. Oracle Spatial has emerged as the DBMS of
choice for the GIS vendors and hence for integration with
photogrammetry, albeit often via an intermediate GIS layer
or component.
In the GI enterprise, the DBMS needs to support all the
processes involved in the care and nurture of the data and in
its delivery to customers. In addition to photogrammetric
workstations it has to support desktop applications, field
update operations, quality assurance processes and delivery
mechanisms. There is some evidence of a trend towards
separate, but linked, maintenance and delivery databases,
because of the markedly different functionality and
performance characteristics of these two regimes (Murray,
2003). This separation can serve to simplify the required
level of integration.
2.2 Joined-Up Data
Another driver towards the database-centric approach is the
need for joined-up data. At its simplest level, this is
expressed as the desire to escape from the arbitrary
constraints of ‘map sheets’ and the associated breaking up of
features in an artificial manner. Even if the system still
operates on a sheet or unit basis, the management of the
resultant edge-matching task is much more feasible in a
database environment. Some mapping agencies (for example
Ordnance Survey Great Britain) have migrated to a
continuous national coverage, with much greater ability to
deliver customised selections of features to users and the
prospect of greater flexibility in internal maintenance. Only
