CityGML – Background and Design - TU...

Institut für Kartographie und Geoinformation, Uni BonnLehrstuhl für Geoinformation

CityGML – Background and Design

Dr. Thomas H. Kolbe

OWS-4 Kickoff MeetingJune 14, 2006

Institut für Kartographie und Geoinformation, Uni BonnLehrstuhl für Geoinformation, Prof. Dr. Lutz Plümer

Dr. Thomas H. Kolbe - CityGML - Standard für 3D-Stadtmodelle 2

Virtual 3D City Models

© Rheinmetall Defence Electronics

© SI, Universität Stuttgart

© T-Mobile

© Sony Corporation

© IKG, Universität Bonn

© Fa. Conterra


OGC Web Feature Service

CityGML

OGC Web 3DService, WTS

VRML / X3D / KML

System A

Stages of Information IntegrationSystem A

GeodataGeodata

graphicalElementsgraphicalElements

System BSystem B

GeodataGeodata

graphicalElementsgraphicalElements

Geodata level

Visualization level

Web Client CWeb Client C


Goals of CityGML (I)

• Establish high degree of semantic (and syntactic) interoperability– enabling multifunctional usage of 3D city models– definition of a common information model– „3D geo base data“ (in the tradition of most European

2D digital landscape models, cadastre models)

• Representation of 3D topography as observed– explicit 3D shapes; mainly surfaces & volumes– identification of most relevant feature types usable

in a wide variety of applications– limited inclusion of functional aspects in base model


Goals of CityGML (II)• Suitability for Spatial Data Infrastructures

– mapping to appropriate exchange format -> GML3• needs high degree of expressivity wrt. OO models• must be usable in the context of OGC Web Services

– possibility to link any CityGML feature to more specialised, functional models / external data sources

• Must be simple to use for applications– well-defined semantics for feature types; however

semantic structure not too fine-grained– subset of GML3 geometries (no curved lines, surfaces)

• Boundary representation with absolute coordinates• advantage: directly manageable within 3D GIS / geo DB


Originator: GDI NRW and the SIG 3D

Geodata Infrastructure North Rhine-Westphalia• PPP founded in 1999 by the state government• more than 100 institutions organized in

6 Special Interest Groups• very active in OGC (GML3, WPOS, Gazetteer, Metadata)

Special Interest Group 3D (SIG 3D)• open group with >70 active members; 3 working groups

– from industry, government, municipalities and academia– participants from all over Germany, Austria, Switzerland, and UK

• aims: interoperable visualization → Web 3D Serviceand exchange of 3D geodata → CityGML


SIG 3D members (excerpt)Municipalities• Berlin• Hamburg• Köln• Düsseldorf• Bremen• Essen• Wuppertal• Bochum• StuttgartAdministration• State mapping

agencies (DE)• Ordn.Survey UK

Companies• T-Mobile• Bayer Industry

Services• Rheinmetall Defence• 3D Geo• CPA Geoinformation• Con Terra• GIStec• GTA Geoinformatik• Inpho• Citygrid (AT)• CyberCity (CH)• Snowflake (UK)

Science• Univ. of Bonn• Univ. of Hamburg• Univ. of Potsdam• Univ. of Hannover• Univ. of Dortmund• Univ. of Münster• Fraunhofer Institute

for Computer Graphics, Darmstadt

• Helmholtz Research Center Karlsruhe

• Univ. of Applied Sciences Stuttgart


Design Principles (I)• Representation of the 4 most important aspects of

virtual 3D city models wrt. multifunctional usage:– Geometry– Topology– Semantics– Appearance

• Usage of topology, appearance should be optional• Semantics restricted to general classifications

– limited inclusion of functional aspects– further functional aspects should be covered by

application specific extensions to CityGML


Design Principles (II)• Adaptibility to different information communities

– values of enumerative attributes are represented in external codelists (increase of semantic & syntactic interoperability)

– communities may redefine the codelists without changing the CityGML schema / data model

• Coherent spatial / semantic modeling– Features are decomposed both wrt. their spatial and

semantic properties

• Balance between simplicity and need for structure

• Multiscale model


Multi-scale modelling: 5 level of details

• LOD 0 – Regional model– 2.5D Digital Terrain Model

• LOD 1 – City / Site model– “block model“ w/o roof structures

• LOD 2 – City / Site model– textured, differenciated roof structures

• LOD 3 – City / Site model– detailed architecture model

• LOD 4 – Interior model– “walkable“ architecture models


Coherent spatial / semantic aggregation

Building

Building installation (dormer)

Building part

Building part

Building

Wall surface (exterior)

3D-Model: City of Coburg


CityGML Dataset in Level of Detail 3

3D-Model: Dr. Benner, Forschungszentrum Karlsruhe


External references

• Every object (part) may have references to corresponding objects in external databases

• Connection with external information, e.g.:– building: link to cadastre, owner's contact information – door, antenna: link to facility management systems

Cadastral DB

Facility ManagementSystem

Telecomm.DB


Terrain Intersection Curve (TIC)

• „Interface between 3D objects and the terrain“– ensure matching of object textures with the DTM– DTM may be locally warped to fit the TIC


Closure Surfaces

• „Seal open 3D objects“– in order to be able to compute their volumes


Thematic Schema: UML DiagramExternalReference

- informationSystem: anyURI- externalReference:

ExternalObjectReferenceType

ComplexRelief loD1GeometryProperty

loD3GeometryProperty

loD2GeometryPropertyGeometry

<<FeatureCollection>>CityModel

_Vegetation

<<Feature>>_CityObject

_Site_Water Bodies

_City Furniture

_Transpor-tation

Objects

CityObjectGroup


Digital Terrain Models

• DTM for each Level of Detail can be composed of – TINs (Triangulated Irregular Network), Grids,

3D Breaklines, and 3D Mass Points– Each DTM component may be restricted to be valid in a

specific region by providing a validity extent polygon

coarseGrid

detailedTIN

Validity extentpolygon can have holeswhich allownested DTMs!


Building Model

• Coherent aggregation of spatial and semantical components– (recursive) composition of building parts– thematic surfaces (roof surface, wall surface, etc.) [from LOD2]– building installations like dormers, stairs, balconies [from LOD2]– openings like doors and windows [from LOD3] – rooms and furniture [in LOD4]

• Components contain relevant thematic attributes– name, class, function, usage, (constr. date, roof type, address)– no. of storeys above / below ground, storey heights

• Surface appearance characteristics (textures, colours)


Building in LoD4 – Interior Model

Entrance door

„Hallway“

„Back room“

„Living room“

Doorway(with door)

Passage(w/o door)

Topology implies an Accessibility GraphRoom


Deriving LOD4 models from IFC

IFC Model

Current research of Benner, Geiger, LeinemannHelmholtz Research Center Karlsruhe


Deriving LOD4 models from IFC

IFC Model

Current research of Benner, Geiger, LeinemannHelmholtz Research Center Karlsruhe

Derived CityGML object


Generic CityObjects and Attributes

• Explicitly modeled feature types have the advantage of well-defined object semantics

• However, often concrete models comprise additional attributes or features not covered by the model

• Incorporation of generic CityObjects and attributes– every CityObject can have an arbitrary number of

generic attributes (string, int, real, date, URI)– GenericCityObject is subclass of CityObject

• arbitrary GML3 geometry for each LOD

– may only be used, if there is no appropriate concept provided by CityGML (problematic wrt. semantic inter.)


Grouping of CityObjects

• Feature type CityObjectGroup– has arbitrary CityObjects as members

• CityObjectGroup is a CityObject– can become again member of another group– every member can denote its role in a group

• usable for user-defined aggregations– e.g. results of classifications or selection

• usable also to group CityObjects wrt. some function or area, e.g. – city districts, building storeys, or evacuation areas


“Backdoor Topology“• How to allow for flexible usage of topology?

– Until now, most 3D city models do consider topology– need to represent city models with geometry only

• Topology model of GML3 sophisticated, but complex– would make it necessary to implement 2 options for

the representation of spatial properties

• Approach in CityGML:– topological connections are represented by Xlinks– GML3 geometries are objects; composites/aggregates

can include subgeometries by value or by reference; – references express topological relations


Geometric-topological Composition

• recursive aggregationarbitrary depth

Solid 1 Solid 2

Solid aggregate

?

SolidGeometry

Solid

SolidAggregate

• Wall face will be partitioned into 2 faces

S.1

S.2

explicit topol. connection- but: goes beyond B-Rep


Semantic Relations by Topology

• Multiple referencing of geometry (components)– realizes topological, but also semantic relations– redundancy free description of space and surfaces possible,

thus no overlaps occur

This surface is part ofthe geometries of the bridge object and the

road object

© Rheinmetall Defence Electronics


Extensions to GML3

• Surface Materials– Colors, Textures (adopted from X3D)– TexturedSurface is subclass of gml:OrientableSurface

• Implicit geometries (Prototypic shapes)– Shape of a 3D object in local coordinates– Instancing at anchor points (+ further transformations)

• Both are concepts used in scene graphs– directly transformable to VRML, X3D, U3D etc.– however only simple & limited extensions– tailored to the demand of 3D city models– easy to support by exporting / importing applications


Prototypic 3D Objects

• 3D city models often contain large numbers of geoobjects of identical shape but at different locations – Examples: trees, traffic lights, street lamps,

benches, etc.

• in GML3, all geometries have absolute coordinates– every copy /instance would have to be

explicitly represented

• CityGML: Implicit Geometries– Separation of shape definition and

georeferencing (anchor point + transform.)


CityGML (Partial) Implementations

• LandXPlorer (3dgeo) • CityServer 3D (GIStec / FhG IGD)• SupportGIS (CPA Geo-Information)• Tridicon (GTA Geoinformatik)• GoPublisher (Snowflake Software UK)• Aristoteles GML3 3D Viewer (IKG Uni Bonn)• Open Source Framework Deegree, WFS 1.1.0 (lat/lon, in

preparation)• 3D Geo DB of the cities Berlin & Bonn;

Oracle Spatial and PostGIS (IKG Uni Bonn, in prep.)


The Official 3D City Model of Berlin

• Background: Need for 3D city model for urban planning, city marketing, and recruiting of new enterprises

• Integration of existing 3D models of the departments of business development and town planning

• Contribution to the SDI Berlin in the field of 3D

• European EFRE Pilot Project from 2003 to 2005

• Currently introduction into regular operation


Realization with interoperable Services

3D City ModelDTM

Orthophotos

3D Geo Database

Versionand History

Management

PresentationSystem

3D City ModelEditor

CityGML

Web FeatureService

Web CoverageService Web Terrain

Service

CityGML

SDI Berlin / Internet


Issues & Items for Discussion

• Generation of CityGML data– few applications available up to now– depends on source data:

• Is it already 3D? (Buildings, DTM)• Are the data entities aggregated to identifiable objects?• Are linear features like streets and rivers represented as

surfaces or by center lines?• Source data formats (Shape 3D, IFC possible)?

• Complexity of CityGML wrt. to current (3D) GIS– multiple geometries per object; topology; multiple

referencing of geometries; LOD model• Integration / Coupling / Relation with IFC

CityGML – Background and Design - TU...

Documents

Transcript of CityGML – Background and Design - TU...