04
n
A REVIEW OF MAP AND SPATIAL DATABASE GENERALIZATION FOR DEVELOPING A GENERALIZATION FRAMEWORK
S. Kazemi*, S. Lim, C. Rizos
School of Surveying & Spatial Information Systems, the University of New South Wales, Sydney, NSW 2052, Australia
PS WG IV/3 Generalization and Data Mining
KEY WORDS: Generalization, Derivative Mapping, Seamless Database, GIS, Map Production, GEODATA
ABSTRACT:
Technological development in the field of map and spatial database
computer-based cartography. Map generalization is an integral part of s
developed and employed by the GIS industry and the computer science
"derivative mapping" from a seamless database as a very active research and development topic. This is an area of interest to many
agencies, academia, map and spatial data providers and users across the spatial industry. It deals with a derivation of smaller sc
detailed single master database. Then the paper provides a brief review of "generalization". This covers the concepts of c
model generalization and generalization operators. It also highlights existing generalization softw
generalize a road network database from GEODATA TOPO-250K. The frame
1:1,000,000 using generalization operators from ArcGIS. The overall
skills in order to derive acceptable results.
1.0 INTRODUCTION
National mapping agencies (NMAs), spatial data providers and map
producers often maintain several databases at different scales, to
represent the geographic world (Lee, 2003; Kazemi, 2003).
Maintaining multiple databases (e.g. small scale to large scale) is
resource-intensive, time consuming and cumbersome (Arnold and
Wight, 2003). In order to serve multiple-purpose and multiple-scale
applications via these databases, automated generalization is a key
solution to be built into modern Geographic Information System
(GIS) software. The aims of this paper are: 1) to highlight the need to
the maintain one master database in order to reduce data handling and
data duplication, and 2) to describe how desired products and
databases should be dynamically developed “on-the-fly” from the
single database using an automated generalization procedure. The
goal is to collect once, and maintain or use at different levels based
on requirements.
Automatic generalization refers to the generation of abstract features
from a rich database through computer algorithms rather than a
human's judgment. It is commonly used for individual objects such
as lines or polygons. Researchers (e.g. Ruas and Plazanet, 1996;
Meng, 1997; Lee, 2002) believe that NMAs and other spatial
information providers/users should work with GIS software
developers to build a universal generalization tool.
The automatic generalization discipline is a fertile research area.
NMASs are committed to maintaining a set of cartographic data with
different scales and to synchronize the updates with other multiple
scale data (Haire, 2001). This is a major challenge for NMAs and
other map/spatial data producers (e.g. Kilpelainen, 1997; Lemarie,
2004). A multi-purpose seamless master database should offer
capabilities to derive different maps at different scales from objects
(e.g. topographic objects), say at scale ranges from 1:250,000 to
1:10,000,000. This capability is referred to as a “derivative
mapping”.
Over the last three decades tremendous efforts have been made to
derive numerical methods (Lee, 2003) applicable to automatic
generalization, in order to generate maps at different scales by
utilizing advanced GIS-based technologies (McMaster and Shea,
1992; Baelia et al., 1995; Joao, 1998). Release of commercial GIS
generalization tools has been well received by major NMAs
(Kilpelainen, 1997; Lee, 2003). Better qualification of generalization
tools in finding reasonable solutions for deriving multiple scale data
from a master database (c.g. Peschier, 1997; McKeown ef al., 1999:
generalization is very fast, following the trend from manual cartography to
patial data collection, representation and access. Most generalization algorithms
community have been tailored for map production. This paper firstly introduces
national mapping
ale map products from a
artographic generalization,
are packages. Finally, it presents a framework to
work will be used to produce small scale maps at 1:500,000 and
aim is to integrate generalization algorithms with cartographer’s intuition and
Thomson and Richardson, 1999; Jiang and Claramunt, 2002) and full
integration of the generalization capability for deriving new datasets
and compiling cartographic products has become inevitable (Lee,
2003). Derivative mapping is composed of several stages, that
include data loading into the generalization software package. A user
needs an identification to adapt and give priority to constraints for
each generalization. Data enrichment refers to the creation of
structural objects such as roads, urban blocks, generalizations of such
objects, and evaluations of the generalization results (Ruas, 2001;
Ruas and Lagrange, 2003).
The remainder of this paper is organized as follows. In Section 2,
differences between the database (model) generalization and the
cartographic generalization in a GIS environment are described. In
Section 3, the relevant literature on generalization operations with
special emphasis on linear features (e.g. roads) is highlighted. In
Section 4, generalization frameworks are reviewed and a conceptual
generalization model is briefly proposed for derivative mapping from
a master database with particular reference to road networks. This is
followed by an overview of generalization software (Section 5).
Finally, Section 6 concludes the paper and indicates research
directions for future work.
2.0 GENERALIZATION THEMES
Weibel and Jones (1998) classified generalization into two main
approaches, known as the cartographer driven (cartographic
generalization) approach and the feature reactive (database or model
generalization or conceptual generalization) approach. This
perspective is revisited here by considering other researchers" points of
view.
Database generalization filters the data through a scale reduction
process, whereas the cartographic generalization deals with
representation or visualization of the data at a required scale (Weibel
and Jones, 1998). A geographic database is usually richer than
cartographic information. The database should offer multiple map
generations in a continually varying range of scales. The latter
method uses an object-oriented data model utilizing data modelling
formalisms to capture the map structure of applications at a given
point in time (Yang and Gold, 1997), since this requires a highly
structured dataset (Brooks, 2000). The object-oriented technology
enables feature definitions and storage as objects with intelligence to
represent natural behaviour of the objects and the spatial
relationships of features. This is based on varying scale in one
representation by displaying certain object dynamically “on-the-fly”
(Zhou er al., 2002; Lee, 1993 and 2002). This type of database can be
Corresponding author: S. Kazemi is currently employed with BAE SYSTEMS Australia. The views expressed in this paper are the author's and not the views of the
BAE SYSTEMS Australia. This study is being carried out as part of a Master's thesis by research in GIS at the University of New South Wales. She can be contacted
via E-mail: Sharon.Kazemi«ga.pov.au.
