VISUALIZATION OF IMAGE QUALITY IN DISTRIBUTED SPATIAL DATABASES
Isolde Schlaisich, Giorgos Mountrakis, Peggy Agouris
Department of Spatial Information Science and Engineering, University of Maine, Orono, ME 04469, USA
(isolde, giorgos, peggy)@spatial.maine.edu
KEY WORDS: Quality, Visualization, Imagery, Databases, Decision Support
ABSTRACT:
Users of GIS are offered an abundance of aerial and satellite imagery in distributed spatial databases. Available datasets can
cover the same area, differing only in quality attributes, such as resolution and positional accuracy. The advantage of having a
bigger pool of information at hand is counterbalanced by the fact that extracting data with the appropriate quality might be
overwhelming. In addition, non-expert GIS users might not be aware of the problems that can arise while working with
datasets of various qualities. Consequently, information about data quality should be provided to users along with the data.
This paper presents a novel approach to communicating image data quality in a visual form, which allows users to grasp the
quality of available datasets at a glance. The data attributes that are identified for holding relevant quality information are
positional accuracy, scale, resolution, completeness, consistency, and currency. To communicate data quality, image quality
information is visualized through 3D models and superimposed color. Delivering quality information alongside original data
helps users choose the data best suited for their task.
KURZFASSUNG:
In ráumlichen Datensammlungen stehen GIS Nutzern eine Fülle von Luft- und Satellitenbildern zur Verfügung. Für viele
Regionen existieren mehrere Datensátze, die sich aber in Qualititsmerkmalen wie Bildauflósung unterscheiden. Dem Vorteil,
den GIS Benutzer durch diese gróssere Auswahl an Daten haben, steht der Nachteil gegenüber, dass sie bei der Auswahl der für
ihre Anwendung geeigneten Daten überfordert sein kónnten. Vor allem Benutzer die keine GIS Experten sind, sind sich zT.
nicht der Probleme bewuflt, die beim Bearbeiten von Datensätzen mit unterschiedlicher Qualität entstehen können. Daher
sollten GIS Nutzern zusammen mit den Originaldaten auch Metadaten zur Verfügung stehen, die die Qualitätsbeurteilung
erleichtern. Dieser Artikel stellt eine neue Methode zur Vermittlung der Qualität von Bilddaten in visueller Form vor. Die
Visualisierung hat den Vorteil, dass der Benutzer die Qualität der Daten auf einen Blick erfassen kann. Die folgenden
Datenattribute werden benutzt, um Qualitätsmaße zu vermitteln: Genauigkeit in der Position, Maßstab, Auflösung,
Vollständigkeit, Konsistenz und Aktualität. Die Daten werden visualisiert mit Hilfe von 3D Modellen und dem Nutzen von
Farben. Wenn Benutzer von räumlichen Bilddaten neben den Daten auch gleichzeitig Zugriff zu Qualitätsinformation haben,
können sie leichter sinnvolle Daten für ihre Anwendung wählen.
The visual channel is the primary sensory input channel
for most people.
Images can convey a lot of information in condensed
space (Beard and Buttenfield, 1999).
People who work with GIS are skilled at absorbing
information from images.
1. INTRODUCTION .
In this day and age users can choose appropriate image data .
for their GIS analysis from a substantial amount of aerial and
satellite imagery. These image collections may be stored in .
distributed spatial databases. In these diverse collections
many datasets cover the same area but have different quality
attributes, offering users a large assortment of information
to choose from, thus attracting a wide range of potential
applications. As applicability of GIS increases,
professionals who are not trained in dealing with geospatial
image data, such as biologists and foresters, discover a
useful tool for their tasks. To help users of varying degrees
of expertise and heterogeneous backgrounds choose useful
images from distributed geospatial libraries, information
about data quality should be provided along with the data.
In this paper we present a novel approach to communicating
image data quality in a visual form, which allows users to
comprehend the quality of available datasets at a glance. As
response to a database query users will get a side-by-side
display of actual data and visualization of the
corresponding data quality. Our environment generates
visualizations dynamically, with parameters and methods
changing according to the query. Visualization, as opposed
to tables of quality specifications, is chosen for the
following three reasons:
The value that data quality visualization adds to a query
system of distributed spatial databases lies in the
combination of spatial and attribute values and their
displaying in map-space, which makes it easier for users to
perceive the data's usefulness and applicability.
As a first step in our approach we had to select which
attributes are most useful for conveying image data quality.
It is rather important to establish balance between showing
attributes that help the user choose the appropriate data, and
at the same time not displaying too many attributes that
would increase the cognitive workload of the user and even
cause confusion. The quality attributes that we chose to
convey the overall image data quality are: positional
accuracy, scale, resolution, completeness, consistency, and
currency. The paper presents arguments for selecting these
specific quality attributes to describe image data quality. To
convey the selected quality attributes we developed
visualizations using representations in 3D and color. 3D
1002
aan A——"—nááákáiBzDS OPQ]QJ] EE ÉL OL]LT
Interne
det oe
repres
height
The re
two €
theory
the u
descri
Sectic
attribi
discus
visual
conclu
In the
resear
shoul
(McG
1997).
lends
param
graph
qualit:
al., 19
Since
to the
The 1
Analy
speart
Qualit
from |
1991)
In the
variou
discus
effort:
210 E
In the
to de:
uncer
fitness
The t
aspect
the fic
error
the w
differ
datab:
Relial
provic
The te
the re
assure
pixels
inforn
bias (1
the te
about
descri
