International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
generalisation. But because the process of generalising data is
time consuming and often hard to define analytically we prefer
a combination of real-time generalisation and using a Multiple
Representation/Resolution Database (MRDB). This database
stores different levels of detail of the same real world objects.
In this study we study especially how an MRDB can be used in
conjunction with the Web Feature Service specification (WFS)
from Open GIS consortium (OGC).
The paper starts with an overview of MRDB. The study
presented in this paper is part of the EU-project GiMoDig. In
section 3.1 a short overview of GiMoDig as well as a
description of the GiMoDig system architecture (which is
mainly based on OGC standards) is given. Then a short
description of the WFS standard follows. Section 5 presents
some case studies. These case studies utilise the system-
architecture of the GiMoDig-service as well as the MRDB-
structure to develop new possibilities to visualise spatial data on
small displays as well as to provide new possibilities to obtain
spatial information. The paper concludes with discussion and
conclusions.
2. MULTIPLE REPRESENTATION DATABASES
2.1 Structure of MRDB
A multi representation database (MRDB) can be described as a
spatial database, which can be used to store the same real-
world-phenomena at different levels of precision, accuracy and
resolution (Devogele et al., 1996; Weibel & Dutton, 1999). It
can be understood both as a multiple representation database
and as a multiple resolution database.
There are two main features that characterise an MRDB:
- Different levels of detail (LoD) are stored in one
database.
- The objects in the different levels are linked.
The first feature can be compared to the analogue map series of
the NMA's: these maps of different scales exist separately, only
implicitly linked by the common geometry. In the second case,
however, individual objects are explicitly linked with each
other and thus each object *knows" its corresponding objects in
the other representations.
Figure 1. Characteristics of an MRDB: Store multiple
representations (left), link corresponding objects
(right).
2.2 Applications of MRDBs
There are several applications of MRDB's. Firstly, they can be
used for multi-scale analysis of the data: Information in one
resolution can be analysed with respect to information given in
another resolution. Gabay and Sester (2002) present an example
where topographic data is linked with cadastral data. A
136
topographic data set of lower resolution containing only
settlement areas is queried concerning the buildings in that area,
information that can be derived from a more detailed cadastral
data set, whose objects are directly linked.
Another application of an MRDB concerns maintenance of
cartographic databases. For example, a major reason for
National Mapping Agencies to investigate and implement an
MRDB is the possibility of propagating updates between the
scales. The appealing idea is that the actual information only
has to be updated in the most detailed data set, this new
information can then be propagated, utilising the links in
MRDB, to all the other scales (Kilpeldinen 1997, Harrie and
Hellstrom 1999).
Vangenot et al. (2002) describe modelling concepts which
support not only the multi resolution view but also the different
views on the object features like object types, attributes and
their values. Kreiter (2002a, 2002b) describes the concept of an
MRDB from the NMA's point of view. Cecconi (2003)
investigates the use of MRDB for the web mapping.
In this study the motivation to introduce an MRDB was to
support and supplement the real-time generalisation. The
benefits of the MRDB are exploited by several other use cases
like introducing adaptive multiscale maps or to give access to
the information of all level of detail stored in the database.
2.3 Combining automated map generalisation and MRDB
To create individual maps for a mobile device real-time
generalisation of the data is often required. Considerable
progress in this field can be observed in recent years (ICA,
2004), resulting in efficient generalisation methods and
algorithms that are applicable to perform scale transitions in
given scale ranges. However, the processes involved going
from a large scale to a small scale (say 1:10k to 1: 1 Mill.) are
very complex. Thus, it is obvious, that (at least today) the
generation and visualisation of ad hoc personalised products of
spatial data in arbitrary scales on a mobile platform cannot be
solved without pre-generalised datasets. ^ Real-time
generalisation can only be efficiently performed in limited scale
ranges and is restricted to operations of minor complexity that
can be solved completely automatically. A way to circumvent
the problem of lack of good generalisation routines is to use an
MRDB.
To minimise the effort of computation work during the real-
time generalisation process, the service selects a scale close to
the desired scale requested by the mobile user. Based on this
neighbouring scale, only limited scale transitions are necessary,
that can be handled in real-time. In this way the need for
complex algorithms, for example displacement, can be
minimised or even excluded.or ;
3. THE EU-PROJECT GIMODIG
3.1 Overview
The EU-project GiMoDig, an acronym for “Geospatial Info-
mobility Service by Real-time Data-Integration and
Generalisation”, aims at developing the spatial data delivery
from national primary geo-databases for mobile use (Sarjakoski
et. al, 2002).
Inte
The
aE
her
bor
hav
ava
dat:
sen
poit
und
3.2
Gilv
data
(ma
sim]
Leh
mak
laye
sect
the 1
a pr
cart
MRI
carte
latte
SVG
Figi
Ther
envir
Unifi
provi
manif
2D sy
Featu
Open
robust
JUME
functi
and al
Durin;
envirc
proces
enviro
