USING IMAGES WITHIN A GIS FOR SPATIAL ANALYSIS
Eugene Derényi, Honorary Research Professor
David Fraser, Research Assistant
Department of Geodesy and Geomatics Engineering
University of New Brunswick
Fredericton, N.B., Canada E3B 5A3
Commission IV, Working Group 1
KEY WORDS: GIS, Database, Image, Raster, Vector, Spatial Analysis, Image Classification, Database Query
ABSTRACT
A universal geographic information system (UGIS), in which both vector based cartographic data and raster based
continuous tone image data can be processed and managed, open new possibilities for performing spatial analysis. The
software tools developed in this environment allows to:
issue location specific database queries through an image display without the superimposition of the vector based
geographic data; and
* extract location specific thematic information from images based either on per-pixel classification or on image
statistics.
The information generated with the help of these analysis tools can directly be entered into the database.
1. INTRODUCTION 2. GEOGRAPHIC DATA
REPRESENTATION
Geographic information is a collection of facts, about a
geographic entity, that has been organized so that it 2.1 Maps
conveys some meaning or significance. The
information associated with a geographic entity has
four major components [Aronoff, 1989]: i) geographic Mapping involves several processing operations, two
location; ii) attribute(s); iii) spatial relationships; iv) of which are data abstraction (or generalization) and data
time. In other words, the four components define reduction. These steps are used to create a model of the
where the geographic entity is, what it is, what its spatial phenomena of interest which is easier to
relationship with its surroundings is, and when was it understand than the real world. As a result, certain
observed. The two most commonly used form of portions of the data gathered of-the-real world are
geographic data representation is the map and the ignored or obscured, while others are. enhanced to
image. There are distinct differences between these two generate a product which meets the requirements of
data representations which impacts on their function users. Nevertheless, maps are useful for portraying
and utilization in spatial analysis. information about the real world because they can
succinctly describe a phenomena, can be easily
During the era when all geographic information was understood and can provide spatial reference when
represented in hardcopy form, the two types of products making decisions regarding a phenomena. Typically, a
were distinctly separated as cartographic line maps and map depicts either a specific spatial characteristic or a
orthophoto maps. In the softcopy world this certain theme. In the digital world the information can
distinction in the two forms of geographic data easily be separated into layers. This approach of
representation exists as digital vector data and raster isolating the real world into layers, is useful for
image data. Advanced geographic information systems examining spatial relationships that exist between
(GISs) allow the co-existence of the two data various themes and the spatial patterns which
representation forms. Thus, the challenge is to exploit characterize these themes.
the advantages of both and to devise innovative
methods for doing so. The spatial analysis scheme Even though maps have proven to be an effective tool
described in this paper is a step in this direction. First, in spatial information management, there are several
however, a short discussion on the characteristics of weaknesses associated with using maps as information
these two forms of data representation is presented. sources, for making decisions regarding the real world.
These problems have been well documented (e.g.
[Unwin, 1981], [Trotter, 1991]) and can be briefly
summarized as follows: Maps are static and often out-
dated even before they are incorporated into a
geographic database. Even though each map is à
generalization of the real world, displaying all map
layers of a multi-layer geographic database at once in
216
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
supei
and :
Anot
gene
to the
dece
inade
2.2
Airb:
syno
vario
varic
acqu
up-tc
They
hand
In Li
two
seco
eyev
inter
prim
Sens:
locat
very
relia
exan
inter
geog
origi
tech
rem
over
usin,
The
with
infor
For
seve
used
[19€
inco
data
[19€
high
date
man
Inte
map
Twc
are |
betv
digi
inte
not
simi
Ima