Developments in 3d Geo-Information Sciences PDF

Lecture Notes in Geoinformation and Cartography
Series Editors: William Cartwright, Georg Gartner, Liqiu Meng,

Michael P. Peterson
Tijs Neutens · Philippe De Maeyer
Editors
Developments in 3D
Geo-Information Sciences
123
Editors
Tijs Neutens Dr. Philippe De Maeyer
University of Ghent University of Ghent
Dept. Geography Dept. Geography
Krijgslaan 281 Krijgslaan 281
9000 Gent 9000 Gent
Belgium Belgium
Tijs.Neutens@ugent.be Philippe.Demaeyer@ugent.be
ISSN 1863-2246 e-ISSN 1863-2351

ISBN 978-3-642-04790-9 e-ISBN 978-3-642-04791-6
DOI 10.1007/978-3-642-04791-6
Springer Heidelberg Dordrecht London New York
Library of Congress Control Number: 2009937575
c Springer-Verlag Berlin Heidelberg 2010

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9,
1965, in its current version, and permission for use must always be obtained from Springer. Violations are
liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws
and regulations and therefore free for general use.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Preface
Realistically representing our three-dimensional world has been the subject of

many (philosophical) discussions since ancient times. While the recognition of
the globular shape of the Earth goes back to Pythagoras’ statements of the
sixth century B.C., the two-dimensional, circular depiction of the Earth’s
surface has remained prevailing and also dominated the art of painting until
the late Middle Ages. Given the immature technological means, objects on the
Earth’s surface were often represented in academic and technical disciplines
by two-dimensional cross-sections oriented along combinations of three
mutually perpendicular directions. As soon as computer science evolved,
scientists have steadily been improving the three-dimensional representation of
the Earth and developed techniques to analyze the many natural processes and
phenomena taking part on its surface. Both computer aided design (CAD) and
geographical information systems (GIS) have been developed in parallel
during the last three decades. While the former concentrates more on the
detailed design of geometric models of object shapes, the latter emphasizes the
topological relationships between geographical objects and analysis of spatial
patterns. Nonetheless, this distinction has become increasingly blurred and
both approaches have been integrated into commercial software packages.
In recent years, an active line of inquiry has emerged along the junctures
of CAD and GIS, viz. 3D geoinformation science. Studies along this line have
recently made significant inroads in terms of 3D modeling and data
acquisition. Complex geometries and associated topological models have been
devised to approximate three-dimensional reality including voxels,
polyhedrons, constructive solid geometry (CSG), boundary representation (B-
rep) and tetrahedral networks. As input for these models, new technologies to
collect three-dimensional data have become fully operational such as mobile
mapping and 3D laserscanning. However, in light of these advances, up until
now there is still a pressing need for robust 3D analysis and simulation tools
that can be applied effectively in a wide range of fields such as urban
planning, archaeology, landscape architecture, cartography, risk management
etc.
vi
In response to the lingering demand for 3D analysis and simulation tools,

a workshop on 3D geoinformation was held in Ghent, Belgium on November
4-5, 2009. Following the successful series of past workshops, the Fourth
International Workshop on 3D Geoinformation offers an international forum
to promote high-quality research, discuss the latest developments and
stimulate the dialogue between academics and practitioners with respect to 3D
geoinformation, acquisition, modeling, analysis, management, visualization
and technology.
This book contains a selection of full-papers that were presented at the
workshop. The selection was based on extensive peer-review by members of
the Program Committee. Only the most significant and timely contributions
are included in this book. Selected contributors were asked to submit a revised
version of their paper based on the reviewers’ comments. All other papers and
extended abstracts that were selected for oral or poster presentation at the
workshop are published in a separate proceedings book.
The editors of this book would like to thank the many people who helped
making this year’s 3D GeoInfo workshop a success. We owe special thanks to
Marijke De Ryck, Dominique Godfroid and Helga Vermeulen for their great
help in organizing the conference, and Bart De Wit and Lander Bral for their
excellent technological support. Thanks also go to Sisi Zlatanova for sharing
experiences and advice on various aspects regarding the workshop, Agata
Oelschlaeger for guiding us through the publication process and our sponsors
for financial support. Finally, we would like to thank the members of the
Program Committee for carefully reviewing the full papers and all those who
submitted their work and participated in 3D GeoInfo 2009.
Ghent, Belgium
August 2009
Tijs Neutens
Philippe De Maeyer
Program Co-chairs
Programme co-chair Philippe De Maeyer

Ghent University (Belgium)
Programme co-chair Tijs Neutens

Local Committee
Marijke De Ryck, Dominique Godfroid, Helga Vermeulen

Program Committee
Alias Abdul-Rahman, University of Technology Malaysia (Malaysia)

Roland Billen, University of Liege (Belgium)
Lars Bodum, Aalborg University (Denmark)
Peter Bogaert, Geo-Invent (Belgium)
Arnold Bregt, Wageningen University and Research Centre (The Nether-
lands)
Volker Coors, University of Applied Sciences Stuttgart (Germany)
Klaas Jan De Kraker, TNO (The Netherlands)
Alain De Wulf, Ghent University (Belgium)
Claire Ellul, University college London (United Kingdom)
Robert Fencik, Slovak University of Technology (Slovakia)
Andrew Frank, TU Wien (Austria)
Georg Gartner, TU Wien (Austria)
Christopher Gold, University of Glamorgan (United Kingdom)
Muki Haklay, University College London (United Kingdom)
Thomas Kolbe, Technical University Berlin (Germany)
Jan-Menno Kraak, ITC (The Netherlands)
Mei-Po Kwan, Ohio State University (USA)
Hugo Ledoux, Delft University of Technology (The Netherlands)
Jiyeong Lee, University of Seoul, (South Korea)
Ki-Joune Li, Pusan National University (South Korea)
Twan Maintz, Utrecht University (The Netherlands)
viii
Mario Matthys, University College Science and Art (Belgium)

Martien Molenaar, ITC Enschede (The Netherlands)
Stephan Nebiker, Fachhochschule Nordwestschweiz (Switzerland)
András Osskó, FIG/Budapest Land Office (Hungary)
Norbert Pfeifer, TU Wien (Austria)
Carl Reed, Open Geospatial Consortium (USA)
Massimo Rumor, University of Padova (Italy)
Mario Santana, K.U. Leuven (Belgium)
Aidan Slingsby, City University London (United Kingdom)
Uwe Stilla, Technical University of Munich (Germany)
Jantien Stoter, ITC Enschede (The Netherlands)
Rod Thompson, Queensland Government (Australia)
Marc Van Kreveld, Utrecht University (The Netherlands)
Peter Van Oosterom, Delft University of Technology (The Netherlands)
Nico Van de Weghe, Ghent University (Belgium)
George Vosselman, ITC Enschede (The Netherlands)
Peter Widmayer, ETH Zürich (Switzerland)
Peter Woodsford, 1Spatial and Snowflake Software (United Kingdom)
Alexander Zipf, University of Applied Sciences FH Mainz (Germany)
Sisi Zlatanova, Delft University of Technology (The Netherlands)
Eric Janssens-Coron, Jacyn
Alfonso Rivera ...................
Data validation in a 3D cadastre

Sudarshan Karki, Rod Thom
From Three-Dimensional Topolo

Yohei Kurata .....................
Needs and potential of 3D city inf

technologies for vehicle positioni
Marc-Oliver Löwner, Andre
Modeling Visibility through Visu

Geo-DBMS
Ivin Amri Musliman, Behna
and Alias Abdul-Rahman...
A 3D inclusion test on large datas

Kristien Ooms, Philippe De
Katharina Bruhm
Dresden University of
Technology, Germany
Manfred Buchroithner
Dresden University of
Technology, Germany
Tet-Khuan Chen
Dept. of Geoinformatics,
Faculty of Geoinformation
Science & Engineering,
Universiti Teknologi
Malaysia, 81310 Skudai,
Johor, Malaysia
Umit Isikdag
Yohei Kurata
SFB/TR8 Spatial Cognition,
Universität Bremen
Postfach 330 440, 28334 Bre-
men, Germany
Marc-Oliver Löwner
Institut für Geodäsie und Pho
togrammetrie, Technische Un
iversität Braunschweig.
Kevin McDougall
University of Southern
Queensland, Australia
Malaysia,
81310 UTM Skudai,
Johor Bahru, Malaysi
Izham Mohamad Yusoff

Department of
Geoinformatics,
Faculty of Geoinformation
Science and Engineering,
University Technology
Malaysia,
81310 UTM Skudai,
Johor Bahru, Malaysi
Sisi Zlatanova
construction methods for 3D models. T
scribed with particular reference to the
neous maintenance of both the primal
complexes in 2D or 3D. They are gene
interiors are possible. All cells are top
rectly with pointers. Our ideas may be
desired, for example for path planning
change of the structure is essential.
Keywords: 3D Data Models, 3D Da

Emergency Response, Disaster Manag
Dual Graph, 3D Graph navigation, Eu
1. Introduction
A cell complex may be considered

ric coordinates at the vertices. A g
the dual structure to a geometric m
T. Neutens, P. De Maeyer (eds.), Developme

Lecture Notes in Geoinformation and Cartog
© Springer-Verlag Berlin Heidelberg 2010
points to the matching Half-Edge,
unique dual HE of the current HE
uses the QE for the shell around ea
replaces the QE Face pointer by th
the “Dual”. The AQE allows direc
the primal or dual graph.
Model construction is difficult u
ient atomic element that can be us
construction operators are difficu
ronoi/Delaunay model. In order t
we developed the Dual Half-Edge
permanent link between the pair
connection of pairs of HEs during
halves the storage with respect to
of low-level construction operator
board-and-tape” construction meth
These higher level operators may
building software, as in CAD syst
served. Our objective is to use this
ning of escape routes from compl
prepared, escape along the edges o
do not provide for management of
Quad-Edge The Quad-Edge (

[5]. Each QE consists of 3 pointe
nected together in a loop by the R
are connected in an anticlockwise
to the next edge with the same sha
nected by this pointer form a loo
pointers R and N are directly use
operators. Rot uses R and returns t
Rot twice. Next uses N and returns
V pointer is used to point to coordi
Fig. 1. Quad-Edge structure and navig
ing R pointer create an edge P0P1; c)
tions between edges in a mesh – q (bla
an example of navigation.
To construct and modify graph

Edge to create a new primal/dual
it to/from the graph under constru
erators could easily be developed
winged-edges [9].
The QE structure was originally
graphs simultaneously. The partic
showing both the primal Delaunay
tion. Either graph may be navigat
and Next pointers, with “Vertex” p
Based on the original Quad-Ed
current full 3D volumetric Euler O
Fig. 2. The AQE structure is suitable f
navigation inside one cell as well as be
As seen in Fig. 2, “Through” p

“Sym” then “Through” again retu
primal space. “Adjacent” is a comp
in the adjacent cell. Thus q.Adjac
sult is the adjacent face from the n
stricted to a single cell; the AQE m
The target application for the AQ
The difficulty with this model was
plex, in particular the “Through”
struction.
2.2. Dual Half-Edge (DH
The previous Augmented Quad-Ed

Stolfi’s [5] Quad-Edge structure. S
Dual Half-Edge structure (Fig. 3)
ist, in order to facilitate construct
pointers must be paired to give the
edge is formed with one or two di
to point to itself or to the second
combinations, as shown in Fig. 4.
different construction methods: Q
and Euler Operators (EO).
The atomic element P0P1D0, for
element used in the QE structure [
space, but the dual edge connects t
that the two distinct vertices are in
ments in construction of 2D mesh
The dual is constructed simultane
2D cells on the 2-manifold. This
CAD systems, but cannot be used f
In [10] we described a spatial mod
spatial entity between. We calle
model, as this is intuitively how it
board”) we start from one vertex
vertices – we split an existing edge
is constructed automatically as the
be used to construct a cell comple
get all points, edges or faces of a
graph, as we would with a single s
by the dual structure, which is also
adjacent shell.
This new idea of keeping the e
described yet (P0P1D0D1 and P0D0 f
Operator spatial model for full 3D
For complete cell complexes the re
but it also allows incomplete item
during the construction process or
new data structure for use in CAD
of non-manifold structure, similar
models.
The main difference between C
struction. Only the EO method allo
model at any time. In C&T we can
the existing model. Another diffe
Two shells (internal and externa
method is used. Even a single ed
make any difference if the created
the external cell is separated when
last face of a tetrahedron, cube or a
two methods is the same (number o
manipulate individual edges as in t
complete faces described above fo
found that we need to work with d
out. Two P0P1D0D1 elements of Fig
Fig. 7. These are then joined to gi
and exterior triangle in 3D space,
process may be developed using P0
The dual to an internal or extern
ternal node. In fig. 4 dual nodes
edges) are not connected directly
In fact there are only two dual no
process – one corresponds to the in
to the external triangle. These two
triangle is not constructed complet
Fig. 7. Construction of a triangle in 3
ternal and external primal shells (black
Using standard Euler operators

cept shells with holes, which are n
used to create a cube from indivi
possible sequences.) For clarity the
step, as with the external shell. T
cells together into one cell comp
closure of the edges, and volume
faces is completed.
structures. We extend the set of st
modifying not only a single shell b
connect/disconnect or merge/split
idea. In fact we perform a sequen
same as we use with standard Eul
ample using standard Euler Opera
cells.
Fig. 9. Merge/Split and Connect/Dis

plex.
Fig. 10. Boundary intersection modul
a bigger face create a new adjacent fac
4. Emergency managem
building models
Geometry, topology and informati

[14]. Geometry and topology are i
be changed using our operators. I
tributes assigned to entities (i.e. v
structure is a pointer based structu
tional attributes. Because of the du
node correspond with a dual volu
treated as if they are connected wi
in various ways, depending on the
as weights for graph traversal, or to
We have shown this model befo
ent data structure with Cardboard
same, but by using Euler operator
are more universal. They could be
objects with included topology.
rooms, d) graph of accessible connecti
5. Conclusions
The DHE data structure and met

models may be useful where both
cell complexes. Models are represe
management and modelling of bu
tions for our models. The geometr
and the topology (connections bet
and the Dijkstra graph traversal a
buildings can be implemented to
amples, where the duality of grap
solve the discrete Maxwell equati
of the dual graph in CAD systems
ber 2008, personal communication
vol. 4 (3), pp. 275-324 (1994)
8 Dobkin, D. P., Laszlo, M. J.: Primi
subdivisions. Algorithmica, vol. 4
9 Tse, R. O. C., Gold, C. M.: TIN Me
Future Generation Computer sys
1184 (2004)
10 Boguslawski, P., Gold, C.: Constr
Dual Navigation Structures, in: L
phy: 3d Geo-Information Scienc
Springer, p. 47-59 (2009)
11 Lee, K.: Principles of CAD/CAM/
ing (1999)
12 Weiler, K.: The Radial Edge Stru
manifold Geometric Boundary M
plications, Elsevier Science (1988
13 Braid, I. C., Hillyard, R. C., Strou
geometric modelling, in: Mathem
sign, ed. K W. Brodlie, Academia
14 Stroud, I.: Boundary Representatio
15 Sazanov, I., Hassan, O., Morgan,
Delaunay Dual Diagram for Co-V
tional Symposium on Voronoi Dia
2007), pp. 199-204 (2007)
visualizations of the respective glac
aerial photographs (2003 – 2006) w
Joanneum Research Graz. By mean
9.1, ESRI ArcGIS 9.2, 3D Nature Vi
treme and Avaron Tucan 7.2 the glac
1850) and the following two dates w
vates of these data sets were generat
First, three individual overflight sim

a realistic impression of the Dachst
2002. As a second embodiment pr
the glacier recession was generated
Third, combining both the flip effec
lar foils, were applied to produce a
play. Fourth, the overflight simulatio
which could then be displayed on
polarization glasses or active shutter
T. Neutens, P. De Maeyer (eds.), Developm

Lecture Notes in Geoinformation and Carto
Fig. 1. Map of Austria in 1 : 2.800.0
Based on historical and recent m

al photography the glacier states
put for photorealistic reconstruc
conditions. A detailed digital terr
2006) were provided by the Gove
University of Graz, Austria (Kro
ping of all major Dachstein glacie
to give an impressive picture of t
1850 and their subsequent retreat
At the beginning of the 20th centu

(Alpenverein) began to explore t
Map” published in 1915 contains
logical conditions. It reveals an e
the glacier area within the past 6
receded above 2000 meter asl, im
their length. The most significan
Gosau Glacier from the Northe
there area were lost (cf. Figure 2)
Torstein Glacier, 3: Little Gosau Gla
Since 1946 the Alpine Club unde

the glacier tongues and frequent s
map was published in 2005 dis
1850 the appearance of the area h
% of the glacier surfaces got los
ing at an altitude higher than 215
going on (cf. Table 1, Krobath &
to import various GIS or geospat
several images formats (Tiff, JP
lected geometric reference system
the terrain data were loaded as a
air photo mosaic. Based on the
the publication of Krobath & L
1850 the necessary vector infor
Maps represent the most accurat
years.
Subsequently the imported vecto

one hand they were utilized to a
vector polygons. Images containi
of grey values) increase or decre
the help of the special tool area t
es between the glacier models
With the help of the Erdas Ima
Fig. 4. Dachstein glaciers in 2002 –
by means of Tucan, also shows a
over the last 150 years.
The Tucan software was used for

puted in real time. By means of t
input data were converted into co
Both free navigation in a scene a

are possible. Besides a free viewe
a professional variant is availab
supported.
The software was designed in su

users without Virtual Reality ex
can offers many functions for th
tions. Based on an integrated Inte
ing geometries as well as dynam
Fig. 5. Screenshot showing the Tuca
Lenticular display
Third, both the flip effect and th

foils, were combined to produce
py display. Lenticular foils offer
processes in the form portable ha
The basic principle of the lentic

viewing created with the help
running on the top side of a tran
interlaced images are arranged w
lenticular display
Fig. 6. Principle of lenticular foil

images.
Display possibilities of the lentic

morphing or animation and the st
2D and 3D effects is also possibl
effect is rather complex. The fli
model were needed. The calcula
the distance of the camera abov
rendering these images are taken
by the lenticular software 3DZ E
to-use lenticular and barrier scre
was used to interlace the images
cm.
Fig. 8. Overhead view of 185
verted into an elevation grid. Ev
glaciated terrain amounts to more
Conclusion
All the above mentioned produc

sense of cross-media visualizatio
the scientific and the public com
ternal tests with students already
vidual products for a better unde
treat in the Alps.
To the authors’ knowledge all the

thods have never before been ap
historical – geodata. This premi
proaches of stereo-visualization
from - - to + +.
Software package
Easiness of use
Generation of complex anim
Integration of GIS data
Import/Export
References
Brückl E, Gangl G, Steinhauser P

schermessungen am Dachstein i
Meteorologie und Geodynamik,
Simony F(1885) Ueber die Schwa
Gletscher des Dachsteingebiges
lungen der kais. königl. Geograp
135
Simony F (1889). „Das Dachsteinge
3D Nature (2009). www.3dnature.co
Awaron Tucan (2009) www.awaron
Digi-Art (2009) www.digi-art.de
Planai (2009) www.dachstein.planai

and interaction capabilities of WPV
nesses. To overcome these limitation
tion of the WPVS++, a WPVS exte
information layers for generated im
requesting spatial and thematic infor
WPVS++ clients can implement var
without changing the underlying w
degree of interactivity and is demons
tions.
1. Introduction
Within spatial data infrastructure

as it allows humans to understan
heterogeneous geodata from dist
For this purpose, the Open Geosp
tion organization proposes severa
ta. So far, there is only one wide
vice (WMS) [4], for generating
geovirtual environments (3DGeo

access information about visuall
information for advanced and ass
reby to increase the degree of in
totype implementations of two w
The remainder of this paper is
and analysis approaches for servi
work. Section 3 presents the exte
the extended interaction model
scribes our prototype implementa
tion 7 concludes the paper.
1 The W3DS has discussion status. T

tion and represents the successor
paper [15].
e.g., massive geometry and textur
In contrast, the OGC Web Pe
portrayal service for generating
perspective views of a 3D scene.
lization, regardless of a client’s r
rendering engine of the server. A
WPVS can be used by more sim
ble bandwidth is low. Additiona
we do not rely on a possible driv
3D hardware, i.e., we do not ha
tions. Using a WPVS, 2D views
by simple URLs and can be easi
systems such as web portals. Na
and far from real-time interaction
tenuated by intelligent loading an
ing).
A major difference between W
optimization strategies for render
optimization strategies such as t
mechanisms are implemented at
formation for deriving new vie
techniques for reconstructing geo
Mobile 3D visualization is m
graphics hardware that became a
years. It supports rendering 3D d
major challenge is to transfer geo
this, elaborated data formats and
sion and buffer compression, fra
required to reduce processing loa
2.3. Image Data and Format
For rendering high-quality image

and takes into account per-verte
vertex normal directions, texture
also incorporated by various ren
cialized visual effects, e.g., ob
ghost views. In computer graphi
for retrieving and storing this ad
referred to as additional (i.e., non
color values, it might be not for o
3. Extending the Informa
In this Section, we propose a con

ing “semantic-rich” images, whic
as a basis for enhanced interactio
3.1 Image Layers
Besides color images, we propos

age layers storing information s
each pixel of the image. Fig. 1 s
provide this additional data as se
does not necessarily encode a c
human cognition.
interest.
Color Layer: A color layer con

pixel. According to the current W
can request a transparent backgr
mats that support transparency, e
Depth Layer: A depth layer des

pixel. Depth images can serve fo
points at each pixel of the imag
rendering techniques such as dep
formation represents a major mea
from the same camera position.
Typically, graphics hardware
values z ∈ [0, 1]. Different from
distances to the virtual camera in
metrics defined by the requested
representation abstracts from co
values can be used without additi
from the color components of the
Object Id Layer: An object id

that refers to a scene object. Usin
els that show a specific feature,
tially analyzing the feature. For
multiple images, object ids shou
that, they could be computed from
dataset.
Other Image Layers: In additio

we identified the following image
• Normal Layer: Describes the
pixel. This could be used for
light sources and the adjustm
e.g., for highlighting scene ob
malized normal vectors in cam
ponent (x, y, z) as color compo
• Mask Layer: Contains a value
ject and 0 otherwise. Thus, a m
Image Data Encoding: Encodin
formats means to support the sa
change, and client-side data load
ditionally, using images allows
algorithms. For color data, whic
lossy compression algorithms c
represents a suitable data encod
values for each pixel, lossless i
suggest PNG to encode this imag
that shall contain transparency.
Technically we encode JPEG
ages by 24 or 32 bit per pixel. H
using image formats with higher
48 (64) bit images. Drawbacks
transfer load, and processing tim
er-side.
The current OGC WPVS provide
point-of-interest (POI) and is no
example, specifying a setting tha
cific point in the 3D world dema
ing distance, pitch, yaw, and roll
As various applications could
view frustum, we suggest replac
one that is based on only three
position (POC), the 3D coordina
optional camera up vector (UP)
thermore, parameters for near p
scribe the culling volume used du
sary, e.g., for the generation an
specification of distorted images
parameter by a field-of-view a
(FOVY) and to complement it b
FOVX is not specified, it is derive
sualization styles to apply, and a
be called image context. Adding
WPVS++ image context could be
Style specifications within this
ble only for requested color ima
layers allow for specifying their
parameterized with the category
usage) and mask images could be
cified by GML-id). Using this
GetView operation is modified
era specifications as described in
4. Extending the Interac
We consider the support of intera

requirement of the WPVS++. T
2 Due to possible non-perspective pr
adapted in future, e.g., to Web View

OP GetFeatureInfo (opt.)
IN
ImageContext: ImageContext
Coordinate2D(1,n): Vector2D
ResponseFormat: string
OUT
FeatureInfo(0,n): anyType
OP GetNavigationCamera (opt.)
IN
Coordinate2D(1,n): Vector2D
NavigationType: MoveCamera
ResponseFormat: string
OUT
Camera : Camera
OP GetMeasurementResult (opt.
IN
As described, feature-related
WPVS++ consumer as non-colo
4.2. Requesting Feature Inf
Similar to the WMS, a GetFea

tra information about the geoobje
ation could either provide domai
mation to the consumer. Accord
capabilities, various response fo
names and values in XML forma
tional formats can be supported a
vice request. In the case of struct
WPVS operation GetDescrip
information for specific datasets.
4.3 Navigation based on Pi
Navigation represents a fundame

virtual globes and online map s
mouse and keyboard in a real-tim
can hardly be defined and desc
Because of that, we propose an
tionCamera that can compute
cation for a 2D pixel position (Li
gest a comprehensible navigation
closer to a selected object. This
center of the bounding box of a s
kept constant; finally, the camer
the object (Fig. 3). This ensures t
of sight keeping the probability o
4.4. Measurement based on
Measuring within a requested vie

retrieve information about the sp
lationships, and the overall spatia
should support the measurement
• Path measurement: Compute
between the 3D positions deri
GetFeatureInfo operation. T
face on top of TCP/IP.
For showing the applicability
two client applications, which
processing power, rendering capa
5.2. Lightweight JavaScript
As example of very lightweigh

based client, which can be run in
ins or libraries. This client just
HTML <img/> elements. It p
moving left/right, tilting up/down
each navigation step a new image
Moreover, this client takes ad
for retrieving 3D coordinates for
and for implementing a “move
GetFeatureInfo operation i
about features in the image.
Fig. 4. Java-based interactive web c
layer (upper left); distance measur
depth layer (upper right); moving t
GetNavigationCamera (lower left an
5.3. Java-based Client
A second client application (Fig

processing and rendering; thus, it
va-based and can be integrated in
so allows for a broad range of app
object id layer). The test was pe
dering was performed on a deskt
NV GeForce 7900 GS), the Jav
connected to the network.
Table 1 lists the number of im
er, compress and send, and the c
ing multiple images in one respo
throughput than requesting each
caused by reduced message over
quests. The gain increases for sm
Table 1. Maximum number of pro

color layer, depth layer, and object
within a single request (multipart res
256
Single part response
Multipart response 1
Factor 2
6.3. Comparison to full 3D G
Compared to full 3DGeoVEs, the

tations. Modalities such as audito
However, provided a sufficient b
for applications and systems base
Navigation capabilities mainly
that consume the WPVS++. Sim
vigation based on requesting and
ditional GetPosition operatio
manipulation. Based on depth i
apply, e.g., image-based renderi
nient and close to real-time visua
Regarding object interaction, t
tion for individual objects of th
are limited: For persistent ma
attributes the original data source
ture Service (WFS) [17]. For use
tive styling mechanisms and spec
W3DS, and render and distribute
In future work, we plan to full
to further formalize the extension
further improving the interaction
ractivity. For example, we will
image layers while still using s
will extend the client application
better using the potentials of the
ally, we will investigate the
photorealistic rendering and illu
and ease the cognition of 3D view
References
1. Altmaier A, Kolbe Th (2003) A

Geo-Visualization. In: Fritsch D
2. Capin T, Pulli K, Akenine-Möl
Graphics Research. Computer G
12. Percivall G (ed) (2008) OGC Re
Consortium Inc.
13. Quadt U, Kolbe Th (ed) (2005)
sion Paper, Open Geospatial Co
14. Saito T, Takahashi T (1990) C
SIGGRAPH Computer Graphics
15. Singh RR (ed.) (2001) Web Ter
cussion Paper, Open Geospatial
16. Verlani P, Goswami A, Naraya
Images: Representations and Re
3D Data Processing Visualizatio
17. Vretanos PA (ed) (2005) Web F
Version 1.1.0, Open Geospatial
fore it is important to develop sim
non-specialists to demonstrate
emerging technologies offer a la
solve this problem and greatly
possibilities.
This paper presents an approac
representation of buildings direc
ing this tool the urban planners a
ings into the existing 3D enviro
such as Google Earth. Several al
be easily created, examined and e
ing the proposed tool the citizens
sign evaluation scenario at home
at a municipality and present the
Java-based and uses newly releas

develop and convey an agreed po
Traditionally, during evaluatio
palities use 2D paper maps, CAD
information and 3D physical mo
tives to citizens. Most municipa
spatial plans, but they are in 2D a
sation. Only recently large mun
for presenting interactive digital
(WMS) (Knap and Coors, 2008).
information technology such as
visualisation and 3D formats for
new possibilities for communica
ternatives. Virtual environments
made 3D visualisation of urban fa
Many municipalities and urban
ing these new functionalities. Ho
tioners that this visual informatio
given representation may be misu
presented message or due to dif
the joint decision making proces
in municipalities, the authors w
uniquely focused on the enablin
(of buildings). In addition, it sh
sented research is not focusing
would support the overall public
The research and literature on
tating planning support and publ
ing provides only a few notable
ment) from the literature in the
experienced researcher in the are
ditional and computerized visua
provided a general map for plann
of options that exist for visual
Hudson-Smith et al. (2002) pres
on a UK case study which is foc
tion and online participation, wh
representation of the urban elem
Simulation Center has been carr
participation in projects such as B
tors in urban development and re
The chapter is organised as f
which covers the aim of the res
the role of Google Earth as the ne
communication, and summarises
models within Google Earth. Th
preparing digital models of exist
ground data for the design propos
introduces a new method for ge
the user interactions on Google E
type implementation realized by
with the conclusion section, whic
tions.
Google Earth as tool for v
3D Geospatial Browser is a ne
pable of retrieving geospatial da
Fig. 1. Visualisation of two 3D build
Google Earth has the ability to

of the buildings which are crea
then be viewed by all the users o
Google Earth, are usually created
SketchUp or other 3D modelling
Fig. 2. Representation of 4 buildings
2009)
The figure above presents the

Google Campus. Their represen
documentation (Google Code, 20
illustration is named as Buildin
transparent red polygons. The po
of the building are generated on-
based on the coordinates of the
multiple polygons. This method
sentation of the building element
applications that acquire/transfer
into Google Earth. In this case,
every face of the building -one-b
in the KML code. We refer this
Representation. An extract from
building with multiple-polygons
The users will not notice the
when a building’s geometry is re
ure 2, in LODs similar to CityG
representation is beneficial for la
be visualised. For example, tests
city of Delft have shown drastic r
file when the buildings are repre
However, it would become impo
ing elements by using the first m
alisation of, the building façade
(i.e. in LODs similar to CityGM
the Multi Polygon Representation
</coordinates>
</LinearRing>
</outerBoundaryIs>
</Polygon>
....
....
</Placemark>
Generation of existing bu
To be able to present and disc

model of the existing area shoul
sentation of urban environment
years. Various approaches exist t
tion about an existing building
source(s) and geospatial models
(Tao, 2006). Among all the app
has emerged as the most innovat
to developing automatic algorit
the polygons, 3) selecting the m
4) extruding the 2D polygon (out
and 5) exporting the created buil
extension. The resulting KML
method (as specified in Section 2
Fig. 3. Snapshots of a) filtering laser
3D reconstructed model in ArcGIS a
Earth. (courtesy, Kibria, 2008).
It should be noted that the sam

(AutoDesk), FME (Safe software
provide tools to create scripts wh
tion of this process.
Generation of buildings dire
In an IT supported joint decis

nicipalities it should be possible
tive design proposals in the virtu
proach presented here is provid
Methodology of the Impleme
In technical terms, there are t

and they both share the same sta
generating the geometry of the f
height. Then the building geomet
• Generated and visualised on-th
• Generated and stored in the fir
These two alternative routes ar
applied for interactive discussion
the citizens to prepare their ideas
velopment effort, the first metho
implementation started with a pro
type for testing and validating th
tation. The following section exp
The Prototype
The prototyping effort include

development of the (interaction)
second stage involved the devel
the building geometry on the fly
component for storing the gener
is still in development at the tim
sections will provide details of th
til today.
Google Earth window,
• the data is first handled by the
• then it is transferred to a dynam
is embedded in)
• finally when the Listener of
Form) notices this change(in th
value of this dynamic attribut
the point that the user clicked)
application.
The Windows Form provides
deactivate the Event Listener of
tions of these buttons will be re
deactivation function is needed
user clicks, when the user inter
building is completed.
eventListenerActive = true;}
in the web page, where the Go

Google Earth API adds an Even
instance. Following this, when th
the following function handles th
ordinates(of the location that the
the web page with the function be
function getLatLonAlt(event) {
lat= event.getLatitude();
document.getElementById("mylat").va
These, i) call of functions fro

ii)passing values from the Java S
made it possible to acquire the u
plan) by user interaction in order
The state of the User Interface
in Figure 6. The List Box in the
nates of the (points that depict t
user interaction with the Google E
Building Model” command in th
stage (i.e. when the button is clic
of stories in the building and the
This information is acquired f
onto the Visual Basic.NET func
of the building. In this first versi
sumption (on a conceptual simpl
as a rectangular-shaped one, wh
and every storey has four-sides (s
wbrowser.Document.InvokeScript("oute
wbrowser.Document.InvokeScript("oute
wbrowser.Document.InvokeScript("poly
End Function
The algorithm uses the Multi

generating the geometry of the b
is represented with a polygon, w
outer rings. In order to generate
tions provided by Google Earth A
the different Visual Basic.NET
below demonstrates some of th
within Visual Basic.NET functio
function outerDraw() {
outer = ge.createLinearRing('');}
function outerAddPoint(lat,lon,z ){
outer.getCoordinates().pushLatLngAlt(la
function outerFinalise(){
polygon.setOuterBoundary(outer);}
nical proof-of-concept that it is
ometries within Geospatial Brow
shows an example of building m
tion.
Conclusion
This paper presented a rapid

decision making in evaluation of
the process involve participation
have mostly different point of v
development. The visualisation
fabric and the proposed building
decision making during the desig
sented in this chapter also techni
tial Browsers for enabling a us
demonstrates that the simple 3D
by user interaction with Google
References
Al-Kodmany, K. 2002, Visualization

From Freehand Sketches to Vi
17(2):189-211
Arayici, Y. 2007 An approach for re
laser scanner for built environm
829
CommunityViz 2009. The Web Site
ware (available at http://www.pl
ESC. 2009 The Web Site of Env

http://www.simcenter.org)
Google Code. 2009 KML Tutorial
TR/apis/kml/documentation/km
Google Earth API. 2009 Google
(available at http://code.google.c
Google Code Playground. 2009 (a
playground/?exp=earth)
Sensing Laboratories, Istanbul, T
Pu, S. 2007. Automatic Building m
P.Van Oosterom, S. Zlatanova,
Geoinformation Systems, LNG&
Reitz,T.,Kramer,M., and S.Thum, 2
based Spatial Data View Servi
Conference on 3D Web Technol
Tao, V. 2006. 3D Data Acquisition
Zlatanova & Prosperi ( eds.)
and Opportunities, Taylor & Fra
Zlatanova, S., L. Itard and M. Van D
vironments used to model bui
IBSA-NVL 2008 Event, 9 Octob
tists often exploit 3D geological m
the complexity of underground s
models is time-consuming and it
software and specific expertise,
is not fully visible and necessitat
zation. In order to simplify this p
an expert system to analyze a pa
geological cross-sections - and t
3D geological models. This pap
the development of this new ex
geological modeling. We will try
information that is required, the
of using such a system, the qual
cussion is based on an experimen
logical Survey of Canada by com
one produced by the prototype 3D
Keywords: 3D modeling, geolog

reasoning

In order to help people to eff
information, we propose to coup
produce automatic 3D geologi
analysis is based on several princ
expert system rules, we chose to
to analyze cross-sections and to
few explanations about the proce
sections analysis. Section 3 pres
and their applications relevant to
describe the 3D GeoExpert tool
geological modeling of the stud
(Canada). In section 6, we discu
expert system for 3D geological m
3D GeoExpert is specifically
applied to hydrogeological app
1 GeOlogical CAD with key com

section and map views an
http://www.pdgm.com/gocad-base-m
lings and cross-sections to ill
Geological maps are 2D represen
ping on the ground surface, whi
the structure and the constituent m
A geological cross-section is
along the vertical axe, the sequ
characteristics as they are found
in polls (subsurface geology). It
objects: units and contacts. A geo
individualized for tectonic and/o
tact is a generic term identifying
It has a specific semantic valu
process that has put it in place [10
Any geological object is assoc
tion. Spatial information relates
2 http://geotopo3d.scg.ulaval.ca/
3 http://www.geoide.ulaval.ca/
they can define sets of geologica
gy such as faults, for example.
2.2. Spatial information and
To identify precisely the info

sections, we have conducted sev
Geological Survey of Canada (G
of the objectives of these intervi
metry, topology, graphic semiolo
cross-section and what kind of i
has highlighted the importance o
lations between objects in the u
sections. This observation is also
posed data model (Fig. 2) is bas
19107 [13]. The spatial data mod
formation included in geological
sections (2D) or the whole 3D m
ical cross-sections and 3D models
In this spatial data model, t

(Point) and corresponds to a tripl
Vertex are the 0D topologic prim
contact is represented by a set o
segments of a same contact. Nod
the modeling type used (straight
rectedEdges are the spatial com
contacts such as faults are repres
ation of the spatial component a
rock) is identified by the feature c
can be defined by the set of Direc
tedFaces are also oriented. Refer
universe that may be 2D, the cro
plex3D are the spatial componen
of geological faces or solids havi
It is important to note that this
crete representation of the 2D (c
and not a continuous representati
be filled and made of pixels (2D
geometric modeling strategy ba
was considered optimal, taking i
the geological model with num
ment) widely used in hydrogeolo
2.3. Geological information
Because geological cross-section

tion, we selected GeoSciML, a d
tion [21]. The GeoSciML project
the 17th century are the basic
sections and even 3D digital mod
• The principle of superposition
wards – unless the entire sect
tion”.
• The principle of initial horizon
posited flat”.
• The principle of original cont
nally continuous. For example
were once connected and hav
them”.
• The principle of cross-cutting
younger than any layer it cuts a
These principles allow geolog

of the geological events that hav
served today. They are essentia
consistent models but also to ana
aternary glacial deposits. The se
tion process is that an important
their reasoning is tacit [9]. It is
identify these unwritten concepts
Several previous work in ge
analysis [7] and 3D modeling [1
geological knowledge can be ex
of a conditional statement such
principle of superposition can be
below the geological unit B THE
work by integrating geological
generation expert system using r
knowledge base includes facts (e
and rules. Rules use known facts
terial is sandstone THEN the ma
are triggered and linked by an in
stored in the knowledge base and
user.
Fig. 4. Steps in the process of analys
The important geological prop

tified, with the help of the user,
tem. The system checks that the
with the GeoSciML vocabulary
properties are then associated to
Fig. 5. A construction scheme of a
(B) The resulting graph with the chr
reading, only the geological units are
file can be built only when all
Figure 5, the profile of coastal de
istence of block 8 (the bloc in th
constructed only in the Step 5.
In the case of discontinuous u
rate elements), or units that don’t
(Fig. 6B), it is necessary to use a
to build the missing sections of t
based on the relative age of geo
cannot integrate geological units
of older geological units. If sev
missing section, then the conta
prior to the unit A is selected. Fo
profile of the unit of Sand and G
Till which is immediately prior
east of the Sand unit), then conta
where there is no Till (Fig. 6B to
sections are represented by graph
algorithm to create the missing s
Fig. 6. Two top profiles. (A) Top p
Sand + Gravel unit.
Once the top profile of a unit

surface of the 3D unit is built by
The top surfaces are built accord
building scheme of the 3D mod
new unit penetrates into previou
pologically inconsistent. The exp
sand and gravel. (B) The entire 3D m
5. Experimentation
5.1. Site test
The experimentation was made o

bec, Canada (Fig. 8). An esker is
It is composed of moraine depo
accumulation. The Saint-Mathieu
a few kilometers wide. It is orien
proximately 11km to the west of
source of drinking water of the ci
Fig. 8. Localization of the St-Math
the esker (in gray) and two cross-sec
We chose this site as a test si

quite simple. There are no com
This reduced the amount of know
We limited our study of the Q
Second, in the context of the GEO
ested to develop a 3D geologica
tion and the integration of 3D ge
Such an esker is representative
which are established in porous
was previously studied by the G
GSC experts using the 3D geolog
eled area by the GSC is about 24
was built from cross-sections and
sess the performance of our syst
model.
Fig. 9. Architecture of 3D GeoExper
The rules and inference engine

Win-Prolog 4.7 (Logic Programm
graphic user interface (GUI) allo
user in order to obtain missing d
ties. We used PostgreSQL v8.3
spatial and geological informatio
general information about model
system.
We used three external op
(www.jgraph.com) is a compon
easy graph visualization. We inte
of the graphs of the cross-sectio
free Java graph library that pro
and algorithms. We used it e
Dijkstra’s algorithm. We used al
lization tools and algorithms of
construction of 3D objects. The
ciated to the geometrical objects
asks simple questions to the user
as what is the lithology or phys
the necessary information, 3D G
cross-sections and then builds the
and Fig. 10).
The entire modeling process (
building) takes about twenty mi
mation to be provided by the use
Points, 60 Nodes, 90 DirectedE
geomodel construction (15 Dire
process. In comparison, the man
the same cross-sections that we p
work.
The arrangement of units and
rect although our model is visua
(Fig. 10). Two reasons explain t
now, 3D GeoExpert only exploits
tal elevation model (DEM). The
strained by information coming f
Fig. 10. Comparison between GSC’
made in gOcad. (B) Our model ob
model obtained with two cross-secti
the terrain above the coastal deposits
tified. (D) Superposition of the GS
(gridded surfaces).
sulting 3D model is visually diffe
with gOcad, it constitutes a solid
er information sources such as d
shape of objects and allow obtain
The integration of an expert sy
fer several advantages including
3D models. In addition, it can fa
for users who do not have as ext
gists. These are the two main co
the main difficulties is the identi
needed for the geological analys
tinguished part of our work was
text. During our interviews with
very useful to identify the object
expert, or system expert, need to
sub-contexts in the modeling co
context. The applicative context
model: "A geological model of w
mines the geometrical type of ob
This type of reasoning is based o
tered by the expert system to me
example, for the identification o
structures (e.g. folds). Besides,
(about 60 rules) allowed us to de
However, a more advanced know
graphs, should be needed to build
However these difficulties sev
pable of intervening in all geol
rience, it would be preferable to
cific tasks, thus reducing the v
acquire.
Concerning the quality of the
two points of view. The first one
put data used to build the 3D mo
two parallel geological cross-sec
cross-sections define the limits o
quality of our model we wish to
lings. They would allow a bette
References
[1] Apel M (2005) A 3D geological

Research Abstracts p. 6.
[2] Bédard K (2006) La construction
sation au tutoriel. Master thesis,
[3] Brandel S, Schneider S, Perrin
trand Y (2005) Automatic buildi
Computing and Information Scie
[4] Brookfield M.E (2004) Principle
[5] Bolduc A, Paradis S J, Riverin
opment of a 3D geomodel for
esker, Abitibi, Québec, Canada.
ing, Salt Lake City, Utah
[6] Campy M, Macaire J-J (2003)
stockage dans les environneme
nod
[7] Chiaruttini C, Roberto V, Buso M
niques in geological modeling.
266
and guidelines. Prentice-Hall Inc
[19] Pouliot J, Lachance B, Brisebo
logical modeling: Are GIS or C
Workshop, WG II/5, II/6, IV/1
poral and Multi-Dimensional Da
[20] Schneider S (2002) Pilotage A
Géologiques Surfaciques,” Ph.D
des Mines de St Etienne, St Etie
[21] Sen M, Duffy T (2005) GeoS
Markup Language. Computers &
[22] Sowa J F (1993) Conceptual G
Technical Report for IRDS Co
National Standards Institute, Ne
own ways of storing and validating
cepts. However, validation rules of
another software’s validation enviro
in great detail but sometimes this
processing. Because of the high vol
pected to handle, it is necessary fo
this paper, validation is initially app
validation? why it is necessary to va
scope to the 3D geometry or spatia
takes a novel approach in identifyin
dastral parcel and identifies the criti
within individual parcels and the re
parcels in 2D or 3D. Although it is d
cal validation rules are examined for
1. Introduction
The cadastre deals with land and

dered to be a combination of geo

space in the 3D cadastral contex
validation rules necessary for a
representations of both 2D and 3
risdiction.
1.1. Structure of the paper:
This paper is structured in six ma

based on existing methods, Valid
in 3D cadastral context, Represe
cussions and further research.
Following a brief introduction
thods of constructing 3D objects
dron, the composite solid object,
ty to the 3D cadastre situation in
Section 3 is a short discussion
3D domain. Questions such as wh
when do we validate are discusse
lidation.
This section deals with the var
jects, and the validation requirem
focuses on boundary representat
section discusses other ways of
vide jurisdictions with options to
individual circumstances like le
and proposed usage etc.
2.1. Objects constructed us
The tetrahedron method of gene

employed in 3D GIS and extensi
Rahman and Pilouk (2007). The
3D cadastre using tetrahedrons.
Validation must ensure:
1. Individual tetrahedrons are n
2. Tetrahedrons used to constr
Discussions:
The validation rules to be co

Rahman and Pilouk (2007) are:
• where tetrahedrons are in con
only, along a line, or on a com
allowed.
• where an object has planar fac
the tetrahedronization proces
whether the original planarity
coordinates.
• it may be considered importan
the faces to ensure that they
nearest faces of Fig. 1 clearly s
to validate this).
• where vertices meet, the coord
• there must be no duplicat
tetrahedrons.
Where composite shaped vol
making sure of the coplanarity. G
volumes addition or subtraction
would have to be checked.
2.3. Regular Polytope
A particular example of the con

the “Regular Polytope” (Thomp
2007). This approach uses as the
fined as the intersection of half s
the union of any number of conv
tic of this approach is that any c
gions is valid, greatly simplifyi
quired is that the size and positi
continuity constraints are enforce
in non-complex situations, where
Fig. 3. City model
The extrusion method of gen

simple to build in a commercial
are not topologically encoded, c
3D data extraction is not possib
This method provides a simple w
validation is limited to the geom
attribute of the extruded 3D obje
Validation is the process of ch
defined rules usually before the
tem. Validation takes on differe
and circumstances of the applica
base administrator would be quit
er. While the former’s validation
integrity checks and business rule
cerned about developing strateg
make over the internet and may
common premises for validation
consistency, data within pre-defi
values etc.
In digital cadastre, the need
tions: (i) who owns the particula
of what is owned. Thus, the majo
unambiguous answers to these q
the extent is generally defined, o
ture, the techniques used, the acc
information is tied to the parcel
stral situation is more complex a
gon faces. However, for a trul
would be treated as duplicate po
which can disallow entry of the
roperability issues.
(0, 0, 1
(0, 1, 1)
(0, 1, 0)
Fig. 4. Defining a
The Open Geospatial Consort

zation for Standardization (ISO)
for generic 3D spatial informatio
are examples of such standardiza
4. Validation in 3D cadas
In this section, situations involv

are listed. These are followed by
rules or specifications that wou
tances. In most cases, the 3D volu
but exceptions may arise in the c
work object may extend beyond t
The situations described below
on similarity of conditions. They
1. Internal validity of 3D parce
2. Surface or base parcel.
3. Relationships to other parce
4. Unique geometrical situatio
5. Further processing on the ge
6. Entry level validations
may form a second complete cov
lowed to overlap.
In this section, the relationshi
base layer is explored, with refe
applicable.
Note that the term "base parc
physically below the "non-base
Oosterom (2006), the correct inte
cel is in fact a column of space fr
low the earth surface to some (po
usually represented as the 2D par
the surface. However, it must be
objects exist in 3D space, for exa
Requirements
The relationship between th

validation requirements such as:
1. Whether the 3D parcel is en
work objects.
• Ensure that the extrusion abo

was completely within the 2D p
If a building is considered to
then the mathematically vertical
right angles to the built surface m
• Ensure that the 3D parcel lies

In some countries, there may
ground rights of the 2D parcel.
dered to be a special case of 3D
vary according to the cadastral i
risdiction. The LADM (ISO1915
by defining the boundary of
represents a set of vertical faces
(Fig 5)
that are part of the 2D base pa
If the base 2D parcel contain
with buildings on a parcel, then
there are overlaps or gaps when t
er, in the case where network obj
fined extents, and surface parcel
gap between the two strata. This
ministrative point of view, but r
parcel, with the defined 3D parce
Requirements
Beyond the relationship betw

parcels, there is a need to consid
cels, including:
1. Checks on situations where
2. Checks on the geometry
relationships with neighbou
Discussions:
• Ensure that there are no gaps

3D parcels.
When considering a simple 2D
vious checks would be to find o
(ii) overlaps or gaps among othe
which would not create distinct p
• Ensure that the sum of the ind
was less than or equal to the b
This is a trivial and obvious ch
jects per individual strata is totall
are contained in identifiable strat
• Ensure that the faces betwee

parcels are formed.
This situation can arise in 2D
not connect to the opposite boun
cels cannot be distinguished (see
complete, but neighborhood oper
parcels would be erroneous. This
treated by most GIS or CAD soft
more complex to ensure that the f
Fig. 7.2. 3D representation of s
+∞
Z=0, or
local
d
3D LA
by laye
-∞
Fig. 8. Utility network object sp

Fig. 9. Mixed use of face string and
volumes (ISO 19152 2009). Limina
the fac
• Existing encroachment. If th
and vertical directions (2D a
created (see Figs 10 and 11),
allows such parcels to be reco
Fig. 11. (Left) 3D parcel number 10
hanging the footpath on the corner o
• “Hanging” title. Titles that h

case of the Story Bridge ove
bridge is leased out to permit t
tour for a fee. This title does n
mapped as a cadastral object,
creates a complexity in storin
not tied to any surface cadastra
Requirements:
1. Effect of subdivision of the

versa
2. Consolidation of contiguous
3. Creating a total/partial reser
4. Limiting of vertical extent o
mining areas, owner’s righ
parcels may be more res
surrounding areas)
Discussions:
• Effect of subdivision
If the 2D parcel is subdivided
ground objects differently. The a
also subdivided and it would be
ject is subdivided proportionately
ever, for below ground objects,
on the extent of the limiting depth
• Non Planar Surfaces

These greatly increase the c
where curved surfaces intersect, w
no simple mathematical definiti
proximated by a number of plana
4.6. Entry Level Validations
In addition to the above mention

tances involving geometry whic
completeness.
Continuity of parcel.
A group of parcels in a cadastral

road due to development. This w
maintain each intersected parcels
cadastre. Most of the validation
and subjective with the spatio-
event time itself and thus gove
above. In some problem domains
alent as 4D primitives, however
assumption of changes as atomic
5. Representations of a
by the ISO 19152 LADM
The ISO19152 LADM, as it

defining parcels (known in the sta
1. point spatial unit (section 5.
2. text spatial unit (section 5.2
3. line spatial unit (section 5.3)
4. polygon spatial unit (section
5. topological spatial unit (sect
tion, “Spatial Unit” is used to refe
Fig. 13. Classes LA_FaceS
LA_SpatialSourceDocument fro
5.1. Point based encodings
Very little validation is possible

coincide (see Fig 14), and fall wi
LA_Fa
locationByText
Fig. 15. Usage f
5.3. Line based encodings
The characteristic of the line-b

‘spaghetti’) is that, while validat
as any of the more highly struct
of interchanging and storing data
The polygon-based encoding sto
while 3D spatial units are stored
one ring, and there should be at l
rings. There is no intrinsic const
overlap, or that there are no gaps
stored once only, with links to th
and 18). As such, there is an in-b
of invalid data from being repres
tuation where a complete non-
defined.
spatial unit (described as “limina
2D parcel, but has to be represen
2009).
The amount and completeness of
cadastral data varies according t
of the databases and applications
fines alternative encoding strateg
lidation is achievable, the data ca
The decision as to what valida
registering authority will be mad
by technology. This philosophy
bited from developing a cadastr
pristine data.
As an example, a cadastre ma
text based spatial units, gradually
This form can be shown to be use
pological failings. Further effort
dastral database.
It is significant that all five lev
ISO19152 can be applied in 2D
combination of 2D and 3D parcel
ISO 19152: ISO-TC211 (2009), G
Domain Model (LADM) Comm
tion for Standardization Technic
CEN for voting 10 July 2009.
Kolbe TH (2009) Representing and

In: 3D Geo-Information Science
Ledoux H, Meijers M (2009) Extrud

consistent 3D city models, In: U
Annual 2009, Krek A, Rumor M
Lemmen C, Oosterom P, Uitermark

forming the Land Administra
(LADM as ISO 19152), FIG Wo
Kazar BM, Kothuri R, van Oostero

Three-Dimensional Geometries.
Van Oosterom P, Penninga F, Z
Thompson RJ (2007) Towards a rigo
thesis, Delft University of Techn
otb.nl/3dtopo/documents/RGI-0
Thompson RJ, Van Oosterom P (2

mentation of integrated 2D an
Geoinformation Systems. Van O
E (eds). Springer, Berlin.
lowing the framework of the 9-i
tact relations based on the presen
surface elements of two objects.
strong correspondence to the 9
Making use of this corresponde
tween various combinations of o
this purpose, we first review and
Then, these topological relations
1. Introduction
Qualitative spatial relations are f

rangements of objects in a spac
tualize them. Spatial database co
qualitative spatial relations, amo
tracted much attention due to it
(Shariff, et al. 1998). Egenhofer
types of topological relations betw

ly, we can distinguish several
(cubes, pyramids, etc.). Thus, the
to record the presence or absenc
ments of two objects in differen
lows the frameworks of the 9-in
and the 9+-intersection (Kurata
topological relations based on th
tween the topological parts of tw
our model have a strong corresp
logical relations. Indeed, makin
contact relations between various
The remainder of this paper is
rizes the concepts of the 9- and
Section 3 develops a model of c
logical relations in and discus
er discussions. Section 5 derives
the corresponding sets of topolo
nally, Section 6 concludes with a
ties of the 3×3 intersections betw
dary, and exterior) of two objec
represented in the 9-intersection
proach, topological relations are
of these 3×3 intersections. Accor
rized by the pattern of bitmap-
whose 3×3 cells indicate the emp
in the 9-intersection matrix (Fig.
° °
M , °
°
B
A M ,
Fig. 1. Representations of a topologi
The 9+-intersection extends th

vision of topological parts. For in
Fig. 2. Representations of a topologi
3. Modeling Contact Rel
Contact relations between two ob

the presence or absence of cont
face elements of an object are th
rectly to the object’s exterior. Fo
guish three surface elements; th
The front and back sides corres
while the edge corresponds to th
dies, we may distinguish differe
stance, we can distinguish six sid
this paper does not distinguish t
body has only one surface eleme
logical boundary. For lines, we
Since all elements in the 9
false), the patterns of the 9+-cont
icons (Fig. 3) in which each blac
of the corresponding element in
Fig. 3 the icon’s first column and
respond to the front and back side
C ,
A
Fig. 3. Representations of a contact r
An attention should be paid to

boundary (edge) of a region and
the boundary of a region A conta
if A’s boundary has an interval
dingly, in Fig. 4, we consider tha
later used for deriving contact r
were studied by Zlatanova (2000
identified the sets of topological
objects in , , and , maki
9+-intersection. Table 1 shows
Among these relation sets, this
line-region relations (Fig. 6), line
lations (Fig. 8), region-body rel
(Fig. 9), each of which contains t
Table 1. Numbers of topological rel

gions, and simple bodies in , ,
emptiness patterns of the 9-intersecti
Point-Point 2 2 2
Point-Line 3 3 3
Point-Region – 3 3
Point-Body – – 3
Line-Line 8 33 33
(a) (b) (c) (d
(i) (j) (k) (l
(q) (r) (s) (A
(F) (G) (H) (I

Fig. 6. 31 topological line-region rel
(D) (E) (F) (G) (H)
(M2) (N1) (N2) (O1) (O2)
(S1) (S2) (T1) (T2) (U1)

Fig. 8. 43 topological region-region
• if A’s boundary intersects with
tersects with B’s interior only o
• A’s interior and B’s interior in
sects with B’s closure or B’s bo
By sketching an instance manual
logical relations hold between tw
fied the following topological rel
limited to convex ones:
• all point-body relation (Figs. 5
• 13 line-region relations in Figs
• 11 line-body relations in Figs.
• 11 region-body relations in Fig
• 18 region-region relations in F
• all body-body relations (Figs. 9
4.3. Topological Relations w
We say that an object A pierces

continuously with B’s interior an
region-region relations in Figs. 8B,E
4.4. Topological Relations w
In Figs. 8F-K,P1/2-W1/2, we can f

regions. Actually, any configu
Figs. 8F-K,P1/2-W1/2 have enclose
gion contains the boundary of an
ly, any configuration of the regio
enclosed space(s). The presence
strictions on the corresponding co
5. Deriving Contact Rela
Contact relations have a strong c

as we can imagine from the str
matrix and the 9-/9+-intersection
derived from topological relation
contact relations between a point
contact icons , are derived
between a line and body in ,
from the topological line-body re
contact relations between two bo
topological body-body relations i
Among these contact relations
to convex ones? Section 4.2 foun
in Fig. A,B, the topological line
the topological body-body relati
Naturally, the contact relations de
the relations that hold in convex c
• two contact relations, ,
• five contact relations,
and convex body; and
• two contact relations, ,
5.3. Line-Region Relations
The mapping from topological lin

bit complicated, because the line
region’s either or both sides, in so
we start from the relations betw
Section 4.2 found that there are
hold in convex cases (Figs. 6a-c,f
• the five relations in Figs. 6a,b,
intersect with the line’s interio
relation, respectively (and thu
are derived);
• the seven relations in Figs. 6c
tersects with the line’s interior
lations in which the line’s int
region (and thus 7×2 contact
are de
mapped to two contact relatio
endpoint contacts the region’s
• the three topological relations
contains the line’s interior, ar
) in which the line’s int
the line’s endpoint(s) follows i
• the two topological relations
intersects with both of the line
rior, are mapped to 3×3 contac
) in which the line’s in
sides of the region independen
• the four topological relations i
rior intersects with one of the l
terior, are mapped to 3×2 con
in which the line’s interior con
the line’s one endpoint contac
respectively;
X’s either or both sides entirely c
contacts Y’s -side. Considering
relations between two regions in
relations in (See Appendix fo
When two regions are limited
much simpler. Section 4.2 foun
that hold between two convex r
O1/2). These relations are mapped
• The two relations in Figs. 8a,b
tion, are mapped to one contac
• The six relations in Figs. 8c-f1
tion, are mapped to 2×2 conta
either side of one region cont
tively;
• The four relations in Figs. 8L1
tersection but no interior-interi
relation (e.g., ) in whic
ther side of another region, res
Fig. 12. A conceptual neighborhood
convex regions in
6. Conclusions and Futu
This paper proposed a model o

work as an alternative of topolog
solid. Due to the models’ schem
tions. It is expected that the comp
siderably different from those o
composition rule of topological r
cludes whole Z, then X include
transformed to the composition r
tacts whole Y, then Y cannot cont
Another interesting future top
tions in . In a 2D space we ca
2D contact relations can be use
ments of two streets and, accord
contact relations will be useful fo
of street networks efficiently with
Acknowledgment
This work is supported by D

through the Collaborative Resear
Strategic Project “Spatial Calculi
Feist, M.: On In and On: An Investi
Scenes. Ph.D. Thesis. Northwes
Freksa, C.: Temporal Reasoning Ba
54(1-2), 199-227 (1992)
Kurata, Y.: The 9+-Intersection: A U
cal Relations. In: Cova, T., Mi
(eds.): GIScience 2008, Lecture
181-198. Springer, Berlin/Heide
Kurata, Y., Egenhofer, M.: The 9+-I
a Directed Line Segment and a
on Behavioral Monitoring and I
Technogie-Zentrum Informatik,
Mark, D., Egenhofer, M.: Modeling
Combining Formal Mathematic
tography and Geographical Info
Randell, D., Cui, Z., Cohn, A.: A Sp
In: Nebel, B., Rich, C., Swarou
Knowledge Representation and
San Francisco, CA, USA (1992)
Schneider, M., Behr, T.: Topologica
jects. ACM Transactions on Dat
- if the topological relations
intersection is mapped to 2
either or both sides one reg
region);
• the topological relations in Fig
four are the contact relations
(i.e., ) and two
sides contact B’s different side
• the topological relation in Fig.
which the boundary of each re
(i.e., );
than the previous case, in wh
both sides of another region (i.
cause once we assume that A’
side contacts A’s boundary (fo
ble side-side contacts (since A’
tions, because once we assume
space (four possibilities exist)
tacts (because A’s x-side may
may contact either or both side
tions in Figs. 8R2,V2 are mapp
• the topological relations in Fig
than those in Figs. 8V1/2 are, in
region contact either or both si
• the topological relations in Fi
than those in Figs. 8R1/2 are, i
sides of the clipped region con
independently;
which either side of one region
• the topological relation in Fi
among which four are the rela
tacts either side of B (i.e.,
where one side of A entirely
side partly contacts either or b
side laser scanner. Today there
stable and absolute positioning i
fers from outtakes, SLAM and d
cise during a limited time span.
duce bevel cuts, real 3D façade
extraction algorithm. Our approa
necting the vehicle to a 3D city d
a spatial request is phased to the
car. There, feature extraction and
solute positioning.
1. Introduction
While traffic situation in fast

individual traffic is going to bec
play a more important role in sa
cuss the shortcomings of today’
these. Our approach combines up
tion algorithms on the vehicle.

continuously available to the AD
of the provided information play
fact that this application will not
to some extend, automatically
emergency braking).
Today digital information ab
support the absolute positioning
use of additional sensors. The m
ierte Liegenschaftskarte (ALK) w
schaft der Vermessungsverwalt
Deutschland (AdV) in the early 8
streets it covers detailed geome
and gully covers etc. Geometries
representing a semantically data
formation system that incompl
heights. Therefore the ALK can
tioning system though it is insuff
A true 3D-data model for virtu
terest Group 3D (SIG 3D) of th
Rhine Westphalia, Germany and
This section focuses on probl
localisation technologies. First, a
ing systems is presented, followe
2.1. State-of-the-art of l
Global vehicle positioning is u

tion. The most popular global n
Global Positioning System (GPS
fence. Applying this system the u
racy of 10 meters depending on
cially in urban environments the
dramatically. Fig. 1 summarizes
To overcome drawbacks of G
ters, steering encoders and in som
urement units (IMU) are integra
requires that the entire localizatio
equipment. The basic principle a
Fig. 1: Most important error source
effects and c) atmospheric effects.
In addition to this conventiona

sors a third group of localizatio
botic community. There, visual s
used to extract landmarks (e.g.
frame. These landmarks are trac
user as well as the landmark itse
localization and mapping (SLAM
tively [8], but requires a continuo
2.2. Major shortcoming
SLAM benefits from the fact t

positions is necessary. On the o
relative positioning. The use of
can lead to an absolute positioni
test arrangement a laser scanner
Fig. 2: Example of a 2D localizatio
tion with a horizontal 2D laser scann
data, the right one a cut out from the
In practice, this 2D approach s

usually situated in the bumper of
the laser scanner the resulting sc
residential area would be mainly
infrastructural obstacles. The ob
cluded in any map due to the f
This problem can be solved by
scribed in more detail in chapte
needs to be extended to 3D.
2.3. Need for up-to-date
3D building representation ca
building footprints and height da
in ground plans like the ALK,
floor numbers or airborne laser
when producing CityGML’s Lo
major problem is the incorpora
about the façades. Even the buil
to detect using 2D ground plans.
horizontal information is extrapo
about building heights and other
under roof surfaces can be identi
heights of gangways, which lead
the façade. Fig. 3 illustrates a ty
to the third dimension. While t
gangway, it is only betoken in t
would lead to misinterpretations
of the vehicle.
Fig. 3: Gangway represented left in
only implied by the ALK but no heig
Next to precision, actuality of

ing solution. In fast growing and
of infrastructure changes within o
tion often changes even faster th
made for roof surface geometry
for building façades. Neverthele
traffic signs and other city furn
carry a static database on the veh
velopment of a Web Feature Ser
integrated database to the car.
Fig. 4: Overall vehicle positioning c
A GNSS based initial positio

WFS. The service phrases a spati
base holding information about t
gorithm. This database is a result
cal data sources known for urb
CityGML data if available, the G
information if needed. The geom
WFS. There, a map matching alg
of the vehicle using the geometri
and the data as a-priori informat
companies.
Evaluation of existing data sou
cle positioning approach is the fi
ity in terms of precision, actuality
of recognition. As a result a catal
fining essential properties of data
tioning approach.
None of the data schemas men
directly. While the German cadas
2D these data would need to be u
a 2D footprint of a building, for i
in section 0. The lack of façade
approach. CityGML data on the
terms of semantics and geometr
would offer much information a
interior and roof structures are r
pearance concept, also and is pa
coded in CityGML’s LoD3. Bes
causes needless data transfer cos
are considered here, it seams to b
Finally the service will prepare
ture data to the vehicle. The tran
versal Mobile Telecommunicatio
The implementation specific
needs to be capable to interpret
the vehicle. In the initial state o
imprecise position which might b
no additional information is ava
rithm will be applied to request
ther, the covered route and velo
amount of data this would mean
applied. As the covered route is
ing box would apply for transm
navigation systems the planed ro
Geographic Data Files (GDF) [
more restricted subset of data fo
GDF data is only a planned rou
transmitted have to be queried bu
longer the positioning service run
be performed. Therefore it is c
Fig. 5: 3D data acquisition using a si
In a first step, the laser scann

fore has to be solved. Therefore t
ner mounted at the front bumper
5. The spatial allocation of the
planes leads to a 3D image of th
Fig. 6: Localization system architect
Against the background of a po

system for safety critical applica
the presented approach plays an
vehicle autonomous integrity mo
the fusion algorithm.
Conclusion
We proposed a concept for an

sitioning architecture in urban en
ing 3D city data and car side la
rithms. Data is requested directly
like gangways are not marked in
formation has to make assessabl
rithm.
Our approach is meant to imp
hicle positioning in urban enviro
vice (WFS) that interconnects the
ing a initial GNSS based positio
sent back to the car. This approa
data as it is stored on the data ba
lenge. Although it is planed to i
like the German Automatisierte L
is likely, that additional data of b
day, no sufficient information ab
port our approach directely.
The proposed service may be
infrastructure. Once a communic
base is established, the car may
like density of traffic or workin
subtracting the effective laser sca
from the integrated data base. Th
and Applications, vols. 1 and 2,
ton, DC.
[7] Titterton, D. H., Weston, J. L., 2
ogy, vol. 207, American Institute
[8] Thrun, S., Burgard, W., Fox, D.,
[9] Schwalbe, E., Maas, H.-G., Se
from airborne laser scanner da
cloud projections. ISPRS Works
[10] Brenner, C. (2005): Building re
In Int. Journal of Applied Earth
187-198.
[11] AdV, 2008: Dokumentation zur
lichen Vermessungswesens
online.de/extdeu/broker.jsp (Acc
[12] Vretanos, P., 2004: OpenGIS W
tion Version 1.1.0. O
http://www.opengis.org (Access
[13] International Organization for S
tems - Geographic Data Files (G
landmark-based navigation, whic
for humans and which also plays
navigation systems. In order to p
discuss one of possible solution
through visual landmarks using
erate measurable visual landmar
which can be used in car or ped
bile application). The focus map
(3D buffering from the 3D shor
Geo-DBMS. Detailing to the g
façade, an implementation of dy
techniques for choosing specific
are shortly presented and their
proposed approach by using Stu
provides outlook on ideas for futu

constraints arising from the use o
to download updated information
pute a location based on landmar
This paper addresses the ques

using visual landmarks within G
are no specific guidelines regardi
landmarks, past relevant research
computer vision and robotics as
Se et. al. (2001). Others are focu
currently interested in - e.g. thro
and Richter 2002) and adding la
support (proposed by Golledge 1
Richter 2002) briefly indicates th
having higher interest to the user
dominant way especially in the
mited displays. In 2007, Neis an
landmarks which can be obta
DWITHIN function in WFS filte
used to measure their attractiven
to dynamic pulse function. Then
proach for the 3D analytical anal
marks will be discussed in Sectio
create new 3D buffering functio
ment and discussions are presen
cluded with some future work rem
2. Navigation Using Lan
Modern car navigation system

class cars and are now available
tively complex and mature system
of digital maps, driving direction
instructions (Zhao, 1997). Obv
landmarks together with corresp
right after the tower’) would be
strength or attractiveness of land
visual attraction (e.g. consisting
semantic attraction (cultural and
shop signs) and structural attract
ues leads to a numerical estimatio
Landmark saliency of a fea

attributes but on the distinction
landmark is a relative property.
space and in the communication
Route directions shall provide a
low someone to travel. Studies s
directions preferably at decision p
study has shown that mapped ro
points lead to better guidance, or
out landmarks. Furthermore, diff
were equally effective (Deakin 1
external points of reference point
in a travel network.
3.1.1. Visual Attraction
Landmarks qualify as visually

racteristics such as a sharp contra
spatial location. Our formal mo
measures criteria regarding vis
complexity.
3.1.1.1. Shape Volume
The shape volume of an objec

its contrast to surrounding obje
whose shape areas significantly
surrounding objects such as large
In the trivial case of a regularly
area is calculated by multiplying
In Geo-DBMS, the volume can

PID>1;
The result:
PID VALIDATION_SOLID3D
---------------------
1 TRUE
2 TRUE
3 TRUE
From the result above, it is sho

type combines three solid objec
mark. To determine which landm
on the sum of numbers of retur
Then it will be sorted again by ca
3.2. Dynamic Pulse Functio
One of ongoing research is the

texturing and dynamic pulse fun
According to our algorithm the g
same rows, columns, size, shape
assumed as one layer. By observ
can make a decision on number o
For example in figure 1: we h

we already mentioned for the w
merged to one layer because the
cluster or distance even if they h
distance. Now it’s time to genera
of output image file and frame in
ter generating the frame of outp
have to measure the geometries o
corner of the rectified picture an
Script program and finally runn
number of window layers that w
do need to run the JavaScript pr
XML schema files with repeated
these files, the XML result schem
image file. The size of the pictur
tal camera is (3888 * 2592) 2.85M
2854) 6.93MB with lots of leanin
er area from exposure point an
parking cars. In the other hand th
* 256) 35.4KB without disturbing
Figure 2: From left to right wall t

layer 1 (251*273)35.3KB, window t
texture for layers 3 and 4 (1053
(481*330)45.4KB
Ivin et al. 2007 has designed t
dynamic shortest path calculation
ues need to be inputted before ru
and end (destination) location (se
After calculating the results,
SQL function (3D line buffering)
file (*.wrl) for displaying shortes
3.4. The 3D Buffering
3D buffering zone could be de

ty information of a spatial objec
features or polygon features in
could be considered as one of th
GIS, such analytical tool is usefu
tion, the research deals with the d
fering tool as a framework for
map using 3D GIS approach. T
primitive, i.e. line.
The buffering object will be

them geometrically) based on th
the object model for buffering z
whereas a line consists of points
mains the same structure. After
geospatial primitives, the buffer
formed since the scope of the
model. There are point (for point
lygon), and surface (for polygon)
the geospatial primitives are n
Therefore, the detailed studies ab
discussed later in this chapter.
3.4.1. 3D Line Buffering
Two end nodes together with

Another phrase like polyline, arc
(2000) had mentioned line data a
jects such roads, railroads, canal
ure 4; line is a combination of bo
line between two successive ve
called line segments. Therefore,
z
Figure 7: Method to create line buff
4. The Implementation
Existing DBMS provides a SQ

storage, retrieval, update, and qu
of the existing spatial database
representing geometries. The ben
Spatial supports the object-rela
The object-relational model u
SDO_GEOMETRY and a single
relational model corresponds to a
tation of spatial feature tables in
for geospatial features.
4.2. Modeling 3D object usi
In the Oracle Spatial object-rela

primitive is possible, e.g. polyhe
polygon to construct a 3D spatia
This is due to two reasons: (i)
Oracle Spatial 11g could yield a
The data structure of polyhedron
place, e.g. POLYHEDRON {}.
within the macro of polyhedron,
(Face3)…}. Finally, the vertices
have multiple, disjoint polygons i
CREATE TABLE Solid3D (

ID number(11) not nul
shape mdsys.sdo_geome
INSERT INTO Solid3D (ID,

1 SDO_GEOMETRY(3007, -
NULL, -
NULL, -
SDO_ELEM_INFO_ARRAY (
1, 1003, 1,
16, 1003, 1,
31, 1003, 1,
46, 1003, 1,
61, 1003, 1,
76, 1003, 1),
SDO_ORDINATE_ARRAY (
4,4,0, 4,0,0, 0,0,0
4,0,0, 4,4,0, 4,4,4
4,4,0, 0,4,0, 0,4,4
0,4,0, 0,0,0, 0,0,4
0,0,0, 4,0,0, 4,0,4
0,0,4, 4,0,4, 4,4,4
)));
Focus
Map
Execute:
Calc3DVolume &
validate_SOLID3D
(GEOMETRY) VALID >
Final Visual Landma

Focus Map
End
Figure 8: Process of generating visu

within the SQL terminal. The va
based on the rules for object co
polyhedron, the validation rules
(2003). A polyhedron is valid if;
1. The data structure is stored cor

2. Each polygon is a flat faces;
3. It bounds a single volume;
4. Its faces are simplicity, i.e. edg
5. Vertices are stored as a face is
Each validation function can

above. However, for the sake of
rules into single validation func
defined function and mapped i
statement denotes the dataset w
within DBMS environment is giv
Figure 9: Geometry table for solid
Figure 10: Geometry table for solid
The 3D buffering function (lin

map based on the resulting 3D
which building will be used for
using dynamic pulse function. Th
tion 5.3 (see Figure 12).
Figure 11: 3D Model Stuttgart, © St
The following SQL statement

It denotes the 3D line buffering f
intersected with the 100m focus
from resulting 3D shortest path
outputs are the intersected faces
S
pa
Building ID = 101
Figure 12: Final Visual Landmarks
Figure 12 shows the result o

DBMS approach. By implement
(largest volume, tallest building a
ings in the focus map (obtained f
namic pulse function for generat
objects (windows, doors, corrido
model will result in high quality
better loading time at the end use
of LoD for visualization, the use
and 3 will took a lot of effort an
bigger image file size as far as t
the use of dynamic pulse functio
navigation applications.
Future work will be specifical

city map information and Geo-S
recognition or awareness. Object
tic indexing into huge informatio
fer from the high complexity in t
viding a means to relate the m
environment. Visual object recog
recognition methodologies is an
bile computing services. Due to t
LSI-Universitat Politècnica de C
Arens, C. A. 2003. Modeling 3D spa

tives. Msc thesis, TU Delft, The
Chen TK, Abdul-Rahman A, Zlatano

ical operations for geo-DBMS.
Zlatanova (eds.): Urban and R
2007, Taylor Francis Group, Lon
Chen TK, Abdul-Rahman A, Zlatan

DBMS environment for 3D G
ICCSA 2007, LNCS 4705, Part
C. Brenner and B. Elias, 2003. Extra

Using Existing GIS Databases
grammetric Image Analysis, In
Remote Sensing and Spatial Inf
2003.
Lynch, K., 1960: The Image of the C
Michon, P.-E.; Denis, M., 2001: W

Giving Directions? In: Montello
ture Notes in Computer Science
M. Schulze-Horsel, 2007. Generati

rence – 3D Architecture 2007.
Neis, Pascal and Zipf, Alexander, 2

using OpenLS Services.
bonn.de/karto/LBSTelecarto200
Olivier Stasse, Andrew J. Davison,

Real-time 3D SLAM for Huma
formation.
http://ieeexplore.ieee.org/stamp/
Raubal, M., & Winter, S., 2002. E

Landmarks. In Geographic Info
Vande Velde, L., 2004. Navigation
tion and LBS, Proc. of 11th Wo
pan, 2004.
Werner, S.; Krieg-Brückner, B.; Ma

tial Cognition: The Role of Lan
man and Robot Navigation. In:
formatik '97. Springer, pp. 41-50
Zhao, Y., 1997. Vehicle Location

Boston, London.
Zipf, A. (2002): User-Adaptive M

Tourism. In: K. Woeber, A. Fre
for Information and Communica
Springer, Berlin.
Zipf, A. and Richter, K.F., 2002. Us

oping Smart Applications for M
(Artificial Intelligence). Special
preprocessing phase is essential
tions. In this initial step, the spac
The voxelization procedure devi
tion types, but the general princip
lows fast retrieval of the status (
points in space.
1 Introduction
An inclusion test is an indisp

ment which users often employ t
within another object. The test is
complex operations. As a conseq
ences the overall efficiency of fre
ly true when analyzing large d
process these data would result
methods are required to handle t
repetition of computational intens

The paper is organized as follo
terature overview of the inclusio
cessing phase is also explained in
3D object representations are des
the Constructive Solid Geometry
clusion test, proposed by Gombo
gerova (2001), is described. This
sion test proposed in this pape
inclusion test are presented in de
justments related to the different
– are emphasized. Finally, a com
thods.
However, when a large numbe
each point has to pass separately
computationally prohibitive. A m
more justified in such a case. Thi
tationally intensive part of the m
sults of this initial phase are esse
datasets. During the preprocessin
certain elements which are more
sion method. These elements are
lik 2005; Zalik and Kolingerova
2005), triangles (Feito et al. 199
gons (Li et al. 2007).
Regarding algorithms of the se

processing phase has to be take
the actual inclusion test. In most
a linear time, O(n), with n also th
clusion test, on the other hand, is
as O(logn) or even in a constan
Huang and Shih 1997; Li et al. 20
Fig. 1. : Relation between the proces
without processing
The algorithm proposed in th

Based Containment Algorithm (C
Therefore, the most commonly u
environment are discussed in the
Representation and the Constru
short description of the 2D CBCA
mented by various data structure
world objects (Stoter and Zlata
when considering inclusion tests
tations create ‘empty’ objects. Th
cations: inside the object, outside
The Constructive Solid Geome

objects which are completely ‘fi
cubes, cylinders and spheres – a
objects, using a set of Boolean op
type of model originates from t
when representing real world ob
tives become rather complex. (St
2.3 The 2D CBCA
The 2D CBCA, proposed by Gom

lingerova (2001), will be describ
the basis for the 3D CBCA propo
Kolingerova 2001)
In the second step, the algorith

cated. The status of the corres
about the location of the point. O
border-status, further calculation
executed in detail or rather appro
cells which are for more than 50
all the other cells are labeled ‘out
3 A 3D extension of the
The proposed method is a 3D e

Algorithm, described in the form
dered to be the most suitable bec
tasets, which is also the aim of th
eling types, are stressed. An ov
found in Fig. 3 and Fig. 4 at the e
3.1 Initialization
The first decision which has to be

This initial choice has far going
cing the initialization time, memo
curacy in a significant way. In g
does not have to be the same, bu
stant sizes is preferred. (Ogayar e
Three methods to define the v

has a fixed voxel size which will
flexible method, allows the user
user has control over the acc
processing times. A third metho
dent on the extent of the scene. T
the present objects) is constructe
visualization aspect.
The methods described in the

sumptions. These methods are
Chen 2000; Thon et al. 2004) an
the specific requirements, linke
way.
3.2.1 Voxelization B-Rep
The Boundary-Representation is
As a consequence, the interior o
analysis it is essential that this e
voxels with status ‘inside’. This
selection of a suitable voxelizatio
Not all voxels in the 3D array

side the object. Certain voxels
which complicates the voxelizati
clusion test is sufficient, voxels
The 3D grid, which was deter
the object. Next, a ray is cast para
origin at the center of each cell
plane. Consequently, if the const
only m x n rays are cast. (Thon et
For each ray – with for examp

box, and the corresponding leave
each surface listed in the leave
When an intersection is found, i
This list is ordered according to a
is processed in a 2D way, cell by
(Thon et al. 2004)
The status of every voxel can

different linked lists. The same ru
be applied in this situation: the
and the object (a bounding sur
counted. If this number is odd, th
be employed on a conventional
next paragraphs is an adaptation
Boundary-Representation in suc
structure of the Constructive Sol
on the main principles of the vo
Chen (2000).
The algorithm for the CSG is

steps: a preparation step and the
step is very similar to the prepara
a quadtree is built, based on the b
leave corresponds with a box par
list with primitives (the composi
which have an intersection with t
The subsequent step, the actu

tured in two ways. Both methods
mitives according to the defined
the first method, the complete p
following the structure of the CS
Each primitive is voxelized in
with a B-Rep model. The only di
of the quadtree do not contain th
section with the corresponding b
which are crossed by the ray. T
listed primitive(s) and the ray are
in an ordered linked list.
Next, the status of all the vox

the z-values of their centers with
number of z-values – with a high
– is odd, then the voxel is locat
status of the voxel is 1 (inside) o
tives, linked with the same paren
according to the correct Boolean
The method described above i

memory, since two whole primi
3.3 Inclusion test
The actual inclusion test will ch

exterior of the object. The initial
a large number of elements (vox
element is known. Hence, the a
which voxel the points are locat
which a point p(xp, yp, zp) is locat
mula:
x p − xmin
rv= ; cv=
sizer
The status of the voxel in whi

the status of the point. A problem
border of the object. Two method
guished: an approximate method
based on the work of Zalik and K
when this detailed test is desired.
to each voxel: one bit for the ins
the boundary voxels. This also g
of the boundary voxels: if a bou
this voxel is located for more tha
This extra information require

age, but makes it possible to per
voxel is ‘boundary’, an extra ray
the preprocessing step can be reu
method in the voxelization step
step.
A ray is cast from the point wh

axis. This ray stops when it e
boundary’. To determine the exa
sections between the ray and th
formation is contained in the qua
If this number is odd, the statu
sion test, is the same for the thre
stant for each point. The main in
of the first phase are:
• the number of faces (or vertice
• the number of primitives (only
• the number of voxels;
• the maximal depth of the quad
An increase in most of these

processing time; only for the de
true. The preprocessing phase is
elization step. Allowing a large
that the time needed to finish
processing time related with the
linked with the maximal depth
depth can be noticed. (Thon et al
The weight of these factors on

of object representation (B-Rep
complexity for the three methods
have fewer faces and thus fewe
times.
The actual inclusion tests can

each point. Consequently, very la
amount of time as the calculation
dependent on the number of poin
intersected intersections
by the ray primitives
Next primitive in the CSG
Next primitive
Next ray
Data/
objects
Process
Fig. 3.: Scheme of the procedures fo

Fig. 4.: Scheme of the procedures fo
4 Conclusion and future
The aim of this paper is to devel

tasets on a conventional desktop
ments. An inclusion test with a p
appropriate to accomplish this. T
form inclusion test which can h
tions: the Boundary Representatio
The proposed inclusion test is

tainment Algorithm described by
thod is adapted such that it can h
combined at the locations where
of the cells is outside in all result
object and ‘inside’ otherwise.
Currently, the methods presen

cal way. Their acceptability, in
with using complexity theory. T
ment the methods described abo
the exact values of the processing
theoretical complexity functions.
References
DeMers MN (2000) Unit 4: Analysi

damentals of Geographic Inform
York, pp 181-395
Fang S, Chen H (2000) Hardware a
ics, 24: 433-442
Taylor G (1994) Point in Polygon Te
Thon S, Gesquière G, Raffin R (200
elization Of Polygonal Objects.
78
Walker RJ, Snoeyink J (1999) Prac
sentations of polygons. In: Goo
Engineering and Experimentati
Springer-Verslag, pp 114-123
Wang W, Li J, Wu E (2005) 2D p
layers. Computers & Graphics, 2
Zalik B, Kolingerova I (2001) A c
for large sets of points. Compute
mentum and continuity aspects of overla
proven to be the shortcomings of the sys
tended dimension representing dynamic s
runoff. The inclusion of Volumetric Soft
and practical 3D dynamic modeling of urb
on Kinematic Wave Routing and Green-A
and overland flow volume respectively.
elements are extracted from Digital Elevat
ing the HEC-GeoHMS program and the H
charge volume gave an R2 of 0.88 and a
that VSG provides end-users the prospect
aster management.
Keywords: 3D GIS, Continuity, Green-A
mentum, Visualization and VSG.
1. Introduction
River basins in Malaysia are now

trial, and infrastructural developm
pervious area in a catchment. The
frequent intense rainfalls, the phy
T. Neutens, P. De Maeyer (eds.), Developme
Lecture Notes in Geoinformation and Cartog
sues pertaining to dynamism has en
changes of 3D data models, to ind
lyze the influence of event based p
tation input, soil infiltration capac
stormwater discharge within strea
term dynamic to refer to the mom
the reference of the model compu
tion is based on the equation of op
in the sub-basin channels.
At present, the large variation a
culties of producing general relatio
tures (i.e. stormwater). Lin et al. (2
ronment (VGE) concept based on
dynamic modelling of atmospheri
is carried out by maintaining an
representation of regular volume
types. Beni et al., (2007) proposed
Delaunay tetrahedralization that de
volume and space at the same time
pological network (grid) for nume
2. VSG for Urban Runoff M
2.1. 3D GIS, Momentum and
Current 3D hydrologic methods are

as soft geo-objects. Urban runoff
water or wastewater that does not
nels, by travelling over the surfa
Ward and Trimble, 2004). Signifi
noff modelling can be determined
spatial connection and the influen
mentums of overland flow and op
rated, exceed the depression stora
continually through soil layers and
rating and linking the urban runof
tion, soil, DEMs and channel netw
use of 3D GIS approach.
Green-Ampt method. This enables
by merging soft geo-objects of ove
ver, the method allows interpretat
urban flood management strategie
The combination of visualization t
3D GIS allows querying of the dat
ing a common format, and transfe
tion. By developing urban runoff
provide a replicable, rational and
processes of the overland flow and
tured framework. Therefore, mode
ban runoff process involve simplif
to improve understanding in a syst
2.2. Mathematical Kinematic

Computation
The Kinematic Wave routing m

open channel (b) with inflow to th
as illustrated in Figure 2.1 (War
Fig. 2.1. Simple watershed with kinem
Direct runoff and open channel

overland flow mentioned by Chow
ence of the overland model com
channel flow, which is the moment
In one dimension, the momentum
where Sf = friction slope; So

sure gradient;
= convective acceleration
gradient can be estimated using Ma
The combination of equations (
kinematic-wave approximation of
The Green-Ampt infiltration m

rate of infiltration to measurable so
conductivity and the moisture con
solutions to the Richards equation.
rainfall based on Darcy’s law assu
porous soil as follows:
f = K(
where K is the hydraulic condu

depth of flow ponded at the surfac
front, and L is the depth from the s
3.1. Study Area
The Sungai Pinang basin lies betw

North and Longitude from 100° 14
nang is the main river system in th
sin size of 51 km2 , comprising m
“Air Hitam” and “Paya Terubong”
located at the West Coast of Penin
storms generated by the inter mo
April/May and October/November
May to September) produces less
whilst the North-East Monsoon, fr
heavier rains to the East Coast of t
rawak (USMM, 2000). Penang Is
mate, which is warm and humid t
nual rainfall of more than 2477 mm
60 mm for February and the highe
The Sungai Pinang basin is a high
Fig. 3.1. Location of Sungai Pinang ba
The procedure for linking GIS

namic modeling of urban runoff p
quisition and development of GIS
Cassini-Soldner projection plane
Routing and Green-Ampt model d
sults within the HEC-GeoHMS
processing of all dynamic flow v
modeling for displaying overland
flood areas. The Green-Ampt and
linked into PC-based ArcView G
software to store and display dynam
Fig. 3.2. Schematic diagrams for det
and flooded area.
To estimate the open channel f

thod, each sub-basin is described
flow planes, collector and sub-col
overland flow planes consist of in
tive slope, overland flow roughnes
loss model parameters. The collect
of area drained by channel, repres
dimensions of channel cross sec
Manning’s roughness coefficient. T
channel length, channel shape, di
representative Manning’s roughne
hydrograph.
Routing Method
The key objective of developing

tinuous spreading of urban runoff,
ences of urban components (i.e. b
jects). The distribution of VSG
dynamism by merging volumes of
The dynamic VSG flow is identif
introduced by Tarboton (1997) to d
on slope gradient.
nel designation as quoted in equati
the merging of VSGs (depicted in
based overland flow, open channe
urban stormwater runoff decreases
top soils. Additional textures on V
high and low infiltrated urban stor
flow and open channel flow within
3.4. Computation and 3D Dyn

Channel Flow, Overland Flow
Simulations of VSG are perform

equation (7) and (8), while open
equation (6). The total of open c
Sungai Pinang basin are compute
the infiltrated rainfall volume usin
tember, 2007 with 10 minutes inte
termining 3D dynamic VSG for u
cylinder.
Fig. 3.5. Flow diagram for rendering 3
4. Results and Discussio
4.1. Potential Urban Runoff A
The experiment of determining

Figure 4.1. Approximately 11.59
Most of the urban runoff coverage
bong”, “Air Hitam”, “Air Terjun R
partly in “Gelugur” and “Jelutong
the sub humid to humid regions, w
urban runoff processes based on m
ristics as stated by USGS (2005).
Fig. 4.1. Potential urban runoff area c
within Sungai Pinang basin based on 1
4.2. 3D Dynamic VSG Modeli

Flow Volume and Urban Runo
Approximately 5,114,100 m3 of
the Sungai Pinang basin. The esti
1,197,000 m3. The total of overla
modeled by VSG are estimated at
The results obtained are illustrated
Fig. 4.2. 3D dynamic VSG modeling
lized at (a) 1 hour, (b) 6 hours, (c) 12 h
Figure 4.2 shows the momentu

the open channel flow and overlan
ni-Soldner plane. Continuous inpu
coverage of the overland flow an
downslopes and flat areas. The mo
Wave Routing method, Green-Am
the equidistant Cassini-Soldner pr
are collected and merged into or
overland flow volume, open chan
charge and flooded areas.
impervious areas. This is the main
indicating that the existing rivers a
shifting runoff volumes from high
then verified and compared with o
ure 4.4.
Fig. 4.5. R2 of Modeled Discharge Vo
The comparison of modeled an

nutes interval gives an R2 of 0.88 a
performance indicator of hydrolo
that the 3D dynamic VSG urban r
step for end-users visualizing com
The differential of modeled and
to the DEM, the borders of rainfal
research carried out by Izham et al
jection plane for preserving adjace
to the land surface. Hence, the mo
due to the different characteristic
each 3D spatial object on the basin
nificantly change the resultant ove
as mentioned by Wang et al. (200
ever does not take the water balan
cludes evapotranspiration losses, p
shallow and subsurface flow, whi
computation of modeled and obser
3. Brutsaert, W. (2005). Hydrology
versity Press, (Chapter 11).
4. Chaudhry, H.C. (2008). Open Cha
(Chapter 2).
5. Chow, V.T. (1959). Open Chann
A).
6. Feldman, R. D. (2000). Hydrolog
System Technical Reference Man
(Chapter 8).
7. Izham, M. Y., Muhamad, U. U., A
Visualization for Infiltration Exces
Sciences, Lecture Notes in Geoinfo
8. Lin, H., Zhu, J., Xu, B., Lin, W.,

ment for a Simulation of Air Pollu
Region. 3D Geoinformation Scien
tography. Springer. 3-14.
9. USMM. (2000). Urban Stormwate
partment of Irrigation and Drainag
sia.

Developments in 3d Geo-Information Sciences PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Developments in 3d Geo-Information Sciences PDF

Uploaded by

Copyright:

Available Formats

Lecture Notes in Geoinformation and Cartography

Series Editors: William Cartwright, Georg Gartner, Liqiu Meng,

ISSN 1863-2246 e-ISSN 1863-2351

Library of Congress Control Number: 2009937575

c Springer-Verlag Berlin Heidelberg 2010

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Realistically representing our three-dimensional world has been the subject of

In response to the lingering demand for 3D analysis and simulation tools,

Programme co-chair Philippe De Maeyer

Programme co-chair Tijs Neutens

Marijke De Ryck, Dominique Godfroid, Helga Vermeulen

Alias Abdul-Rahman, University of Technology Malaysia (Malaysia)

Mario Matthys, University College Science and Art (Belgium)

Data validation in a 3D cadastre

From Three-Dimensional Topolo

Needs and potential of 3D city inf

Modeling Visibility through Visu

A 3D inclusion test on large datas

Izham Mohamad Yusoff

Keywords: 3D Data Models, 3D Da

A cell complex may be considered

T. Neutens, P. De Maeyer (eds.), Developme

Quad-Edge The Quad-Edge (

To construct and modify graph

As seen in Fig. 2, “Through” p

2.2. Dual Half-Edge (DH

The previous Augmented Quad-Ed

Using standard Euler operators

Fig. 9. Merge/Split and Connect/Dis

Geometry, topology and informati

The DHE data structure and met

First, three individual overflight sim

T. Neutens, P. De Maeyer (eds.), Developm

Based on historical and recent m

At the beginning of the 20th centu

Since 1946 the Alpine Club unde

Subsequently the imported vecto

The Tucan software was used for

Both free navigation in a scene a

The software was designed in su

Third, both the flip effect and th

The basic principle of the lentic

Fig. 6. Principle of lenticular foil

Display possibilities of the lentic

All the above mentioned produc

To the authors’ knowledge all the

Generation of complex anim

Integration of GIS data

Brückl E, Gangl G, Steinhauser P

Simony F (1889). „Das Dachsteinge

3D Nature (2009). www.3dnature.co

Awaron Tucan (2009) www.awaron

Digi-Art (2009) www.digi-art.de

Planai (2009) www.dachstein.planai

Within spatial data infrastructure

T. Neutens, P. De Maeyer (eds.), Developm

1 The W3DS has discussion status. T

2.3. Image Data and Format

For rendering high-quality image

3. Extending the Informa

In this Section, we propose a con

3.1 Image Layers