You are on page 1of 75

3D-Stadtmodelle fr die Metropolregion Rhein-Neckar?

Potentiale und Erfahrungen

GIScience Research Group


Department of Geography
University of Heidelberg
http://giscience.uni-hd.de
zipf@uni-heidelberg.de

Quo Vadis?
3D-SDI Potentials
Service architecture
Applications

Experiences
Heidelberg 3D
OSM 3D

GIScience
University of Heidelberg

3D City Models

Business Planning

(Pedestrian) Navigation

[Google Earth]

http://giscience.uni-hd.de

GIScience
University of Heidelberg

[virtualcitysystems.de]
3

Simulation models
Air flow, immissions, noise...

Spatial Data Infrastructures (SDI) today

Geodata accessible through the Web based on


open standards (ISO, OGC)
Open Geospatial Consortium (OpenGIS)
Specifications for Web Services & data formats (GML)

Web Catalogue Service (CS-W)


Web Map Service (WMS)
Web Feature Service (WFS)
Web Coverage Service (WCS)

Metadata
Maps
Vector
Raster

Where next?
GIScience
University of Heidelberg

Whats Next my choice:


3D
Dynamic Sensor Data
Processing & Analysis

osm-3d.org

based on different data:


from official sources
private industry or
open crowdsouring

GIScience
University of Heidelberg

gdi-3d.de for Heidelberg-3D.de

GIScience
University of Heidelberg

The big picture: 3D cities on the Web


Today: proprietary virtual globes
- Google Earth / Microsoft Bing
- Nasa Worldwind

Only Visualisation !

OGC working group: 3D Portrayal Services


- Web 3D Service (W3DS)
- Web Perspective View Service (WPVS)
- Symbology Enconding 3D (SE-3D)

Integration in SDI Framework !


GIScience
University of Heidelberg

3D SDI Core Components


Current OGC Standards can be used, but
extend the functionality or the specification

Database(s)
Storage &
maintenance of 3D
data, CityGML Import
-> 3D GeoDB / WFS
Applications...
Routing, sensors, LBS
-> OGC OpenLS
-> OGC WPS
-> OGC Sensor Web

Metadata

Catalogue Services
provide meta data
-> ISO 19115

Web3D Service

Provides 3D
graphics
-> OGC W3DS
Discussion draft

Styling

Enables user
defined styles and
visualization rules
-> OGC SLD -> 3D
SE

3D Viewer

UI & access to
services

OGC Web 3D Service (W3DS)


Support for streaming through tiling &
view dependent level of detail (LOD)

GIScience
University of Heidelberg

W3DS Visualization Server Concept


Web

Visualization Server
- Layer Management
- Styles Management
- LOD Handling

D Service Interface

Export Module

3D DB
Modelling Engine
- LOD Generation
- DGM Triangulation
- Extruder etc.

- VRML
- KML
- X3D

Object Library

Import Module

2D

VRML / CityGML

2.5 DB

GIS 2D Vector Data

Landmarks

DEM Database

- Medium Server, Medium Client


- Client recieves 3D display elements
- Client side 3D rendering
- Server processes raw data and
generates complete 3D scenes

GIScience
University of Heidelberg

CityGML Import
GIScience
University of Heidelberg

OSM-3D W3DS Server & Google Earth


OGC 3D Interoperability Pilot

GIScience
University of Heidelberg

Data Update?
Automated nightly synchronisation with 2D cadastral data in bureau of
surveying Heidelberg through OGC interfaces (WFS-T)

GIScience
University of Heidelberg

Generating 3D landscape models


Integrate heterogeneous data sources:
3D buildings (city model)
digital elevation models (DEM)
triangulated 3D points

Landuse (streets, water, forest, parcels)


2D vector data with landuse type
aerial / satellite images (no semantics)

other objects
trees, signs, street furniture
landmarks, POIS,

GIScience
University of Heidelberg

Level of Details (LoD)

Example LOD-tiles for different scales


Edge Contraction for DEM Simplifications

GIScience
University of Heidelberg

Smoothed streets in 90 meter DEM


Adaptive flattening of streets improves original coarse
DEM (energy function dependend from land use moves triangles)

GIScience
University of Heidelberg

Vector map style versus aerial photo

GIScience
University of Heidelberg

3D Symbology Editor

GIScience
University of Heidelberg

Calculation of emission simulations


dynamically calculated by WPS
with actual wind information
through OGC SOS
(Sensor Observation Service)

only proof of concept

GIScience
University of Heidelberg

From static to dynamic data in 3D SDI


provide real-time sensor data within SDIs
based on OGC Sensor Web Enablement (SWE)
Web-based access and control of sensor networks
Example: Integration of different sensors in 3D-SDI

GIScience
University of Heidelberg

Sensors in GDI-3D

GIScience
University of Heidelberg

building monitoring

WebCam per OGC SOS (Sensor Observation


Service)

GIScience
University of Heidelberg

OpenLS Route Service 3D

Automatic generation of 3D graph; Usage of 3D route service

3D Indoor Routing (OWS-6 testbed)

GIScience
University of Heidelberg

Does it scale to larger regions?

GIScience
University of Heidelberg

Bringing VGI from 2D to 3D: OSM-3D.org

GIScience
University of Heidelberg

OSM-3D.org

GIScience
University of Heidelberg

GIScience
University of Heidelberg

Francfort

Munich

GIScience
University of Heidelberg

OSM-3D: Tokyo, Dubai...

http://giscience.uni-hd.de

GIScience
University of Heidelberg

OSM-3D.org
Only few bottom-up approaches based on VGI so far
OSM-3D: a 3D globe based on OSM & SRTM data

http://giscience.uni-hd.de

GIScience
University of Heidelberg

OSM Buildings
Ca. 98 Mio worldwide

http://giscience.uni-hd.de

GIScience
University of Heidelberg

Building Generation
Several OSM keys contain relevant information for the building
Direct height: height, building:height
Indirect height: levels, building:levels, building:levels:aboveground
Elevation: building:min_level, building:min_height
Roof Shape: building:roof:shape, building:roof:style, building:roof:type
Roof Colour: building:roof:colour, building:roof:material
Facade Colour: building:colour, building:facade:colour, building:cladding
Additional Roof Attributes:
- building:roof:extent
- building:roof:orientation

GIScience
University of Heidelberg

Compiling 3D City Models


Common data capturing and editing process
Aerial/Terrestrial Laser Scanning & Photogrammetry
(semi-)automatic reconstruction, high-quality data
Accomplished by dedicated 3D companies /
surveying administrations

[ikg.uni-hannover.de]

[mobilecadsurveying.co.uk]
[ Geodigital, LiDAR As Art]
http://giscience.uni-hd.de

GIScience
University of Heidelberg

35

3D Building Reconstruction
much research exists about 3D reconstruction of roofs,
facades etc. from high-quality point clouds

[Mller et al., 2007]

[Huang et al., 2011]


http://giscience.uni-hd.de

[Pnard et al., 2005]


GIScience
University of Heidelberg

[Pu & Vosselmann, 2009]

[Sampath & Shan, 2010]


36

Crowdsourcing 3D
Prerequisites for 3D-VGI are better than ever

[ Autodesk / genbeta.com]

low-cost hardware, service-based photogrammetry

[ Microsoft]

Increasing awareness & interest in 3D in our society


Collaborative capturing of 3D geo-information becomes possible

http://giscience.uni-hd.de

GIScience
University of Heidelberg

OpenBuildingModels.uni-hd.de
What?
Free-to-use repository for 3D architectural building models
Link models to OSM or use them for other applications

Why?
a)

Not each and every building is taggable

[gigalo.de]

[detaildesignonline.com]
[wikipedia.org]

[nicetobook.com]

b) It is feasible! There exists already a huge 3D community


cf. Google 3D Warehouse, OpenSceneryX.com, Archive3D.net,
Free3DModels.org, Shapeways.com ..
http://giscience.uni-hd.de

GIScience
University of Heidelberg

38

OpenBuildingModels.uni-hd.de
Geo-referencing
Match with existing OSM groundplan
T t x , t y , sx , s y ,

Local
CRS

(+ height)

y
x

http://giscience.uni-hd.de

Global
CRS

GIScience
University of Heidelberg

OpenBuildingModels.uni-hd.de
Integration into OSM-3D
great leap forward towards collaborative 3D city models
Significant improvement of LoD
Non-taggable buildings are particularly important (landmarks)
Beneficial for many applications

http://giscience.uni-hd.de

GIScience
University of Heidelberg

40

Vielen Dank

GIScience Research Group


Department of Geography
University of Heidelberg

Crowdsourcing 3D: Future Directions


Broad range of capturing/editing/visualisation tools
Research about appropriate data structures & workflows

http://giscience.uni-hd.de

GIScience
University of Heidelberg

42

Conclusion
Recent VGI projects have shown their potential
New: 3D-VGI, citizens as 3D-sensors
Sensors become cheaper and more available

Main requirements:
Suitable data structures for crowdsourced 3D-modelling
Broad range of capturing and editing tools, viewers

Collaboratively created 3D city models through generative


methods can be enhanced with such a repository of
architectural 3D models

http://giscience.uni-hd.de

GIScience
University of Heidelberg

43

OGC Web 3D Service (W3DS)


delivers 3D scenes
display elements
VRML, X3D, KML, Collada

imports 3D city models


OGC CityGML

defined http interface

http://myserver/w3ds?
request=GetScene&
bbox=3448244,5540223,3448582,5540495&
poi=3448410,5540385,0&
layers=trees,litfass,laterne,ampel,buildings,streets,rhein&
styles=tree,litfass,laterne,ampel,bisque,black,lake
&aov=45&pitch=45&roll=20&yaw=270&distance=80&

..

GIScience
University of Heidelberg

Analysis in 3D on the Web?


Web Processing Service 3D (WPS-3D)
Processing & Analysis of DEM & 3D geodata
DEM (pre)processing: tiling, triangulation, generalization
Domain specific scenarios
bomb finding

(disaster management)

emission spread
DEM analysis
Slope, Aspect

Profile Graph
Visibility / Viewshed
...

GIScience
University of Heidelberg

Server-Client-Balance

Image file
VRML

(OGC 2005)

GIScience
University of Heidelberg

Domain specific workflow: generate 3D city- &


landscape models

Prototype:
Focus maps 2D & 3D

New OGC W3DS Spec draft:


GetTile-Operation
Include definitions of TileMatrix and TileMatrixSet

TileMatrix

TileMatrixSet

GIScience
University of Heidelberg

OpenBuildingModels.uni-hd.de
Current state: prototypical web-platform

Process building outlines from OSM ways


Provide them as a WFS (GeoServer)
User can select building
Upload 3D model to repository

http://giscience.uni-hd.de

GIScience
University of Heidelberg

50

Thank you very much

GIScience
University of Heidelberg

Distributed W3DS with 3D-Sybology


Encoding

GIScience
University of Heidelberg

Integrated DEM: LOD Generation

Selection of required 2D layers (landuse types) for each LOD


Tiling for several scales (LODs)
Scale dependent landuse generalisation
Triangulation of 3D points
Mesh simplification via an edge contraction algorithm
preserving boundaries of the 2D layers

Adaptive DEM modification


of 2D line features (streets, railways...)
Generation of one 3D file
for each LOD, tile and land use type
Generate / place buildings and other objects on DEM
Computing intense: GRID-Computing
GIScience
University of Heidelberg

Generalisation of integrated DEM

LOD 500
1699 KB

LOD 1000
552 KB

LOD 8000
102 KB

LOD 16000
43 KB

GIScience
Example: area of 3600 * 3600 meter
(Heidelberg); file size in kilobyte
University of Heidelberg

Public transport network in OSM3D.org

GIScience
University of Heidelberg

The explosion of data


technical sensors
citizensas sensors
(crowdsourcing, citizen scienc, social media)

http://giscience.uni-hd.de

GIScience
University of Heidelberg

Bringing VGI from 2D to 3D: OSM-3D.org

GIScience
University of Heidelberg

OSM-3D.org

GIScience
University of Heidelberg

OSM-3D: Tokyo, Dubai...

http://giscience.uni-hd.de

GIScience
University of Heidelberg

59

Compiling 3D Cities a new approach


How about Crowdsourcing 3D City Models?
Motivated voluntary mappers have shown their potential to do it!

St. Peters Basilica, Rome

[Google 3D Warehouse]

Disneyland, USA
[OpenStreetMap]
http://giscience.uni-hd.de

GIScience
University of Heidelberg

60

OSM-3D.org
Only few bottom-up approaches based on VGI so far
OSM-3D: a 3D globe based on OSM & SRTM data

http://giscience.uni-hd.de

GIScience
University of Heidelberg

OSM Buildings Worldwide


Ca. 95 Mio worldwide

http://giscience.uni-hd.de

GIScience
University of Heidelberg

Building Generation
Several OSM keys contain relevant information for the building
Direct height: height, building:height
Indirect height: levels, building:levels, building:levels:aboveground
Elevation: building:min_level, building:min_height
Roof Shape: building:roof:shape, building:roof:style, building:roof:type
Roof Colour: building:roof:colour, building:roof:material
Facade Colour: building:colour, building:facade:colour, building:cladding
Additional Roof Attributes:
- building:roof:extent
- building:roof:orientation

GIScience
University of Heidelberg

OSM-3D.org: OSM Landuse & SRTM DGM


,
SRTM DHM

OSM Daten

Kachelung

Selektion

Triangulation

Generalisierung

Generate
buildings

(Douglas-Peucker)

TIN
Integration
OSM 2D Daten

TIN Vereinfachung
(Garland & Heckbert)

Einebnung
linearer 2D Daten

Visualisierung
Integrated DEM GIScience
(VRML) University of Heidelberg

POIs (labels,
signs)

Francfort

GIScience
University of Heidelberg

OSM-3D: Tokyo, Dubai...

http://giscience.uni-hd.de

GIScience
University of Heidelberg

Problems
Mapping for the Renderers
OSMers map buildings mostly for the sake of visualization

Quelle: Bing

http://giscience.uni-hd.de

GIScience
University of Heidelberg

Templates for POI 3D features


There are many point geometries in OSM which describe re-occuring
elements, such as trees or lights, windmillls, traffic ligths, wind power

http://giscience.uni-hd.de

GIScience
University of Heidelberg

osm-3d.org planet

http://giscience.uni-hd.de

OpenLS local search/ Routing in OSM-3D

http://giscience.uni-hd.de

GIScience
University of Heidelberg

OSM-3D W3DS Server & Google Earth


OGC 3D Interoperability Pilot

GIScience
University of Heidelberg

From OSM-3D to CityGML


Initial question:
Is it possible to use VGI of OSM for the creation of 3D CityGML
models
Answer:
Considering the existing buildings in OSM, an automated
creation of CityGML LoD1 (blocks model) and LoD2 (distinction
of building parts such as roof, wall, ground) is feasible. For more
detailed models (especially LoD4 with indoor spaces), OSM
lacks appropriate mapping models.

Goetz, M., Zipf, A. (2012): Towards Defining a Framework for the Automatic
Derivation of 3D CityGML Models from Volunteered Geographic Information.
International Journal of 3-D Information Modeling (IJ3DIM), 1(2). IGI-Global
http://giscience.uni-hd.de

GIScience
University of Heidelberg

72/12

From OSM-3D to IndoorOSM


Question:
How can OSM be extended towards indoors, i.e. how should a
scientifically reasoned and well thought indoor mapping proposal look
like?
Answer:
Based on a comprehensive but simple 3D Building Ontology (3DBO),
an OSM mapping proposal has been developed which on the one way
utilizes existing OSM methodology (nodes, ways, relations, tags) and
on the other hand can already be realized in existing OSM editors

Goetz M., Zipf A. (2011): Extending OpenStreetMap to Indoor Environments: Bringing


Volunteered Geographic Information to the Next Level. In: Rumor, M., Zlatanova, S.,
LeDoux, H. (eds.) Urban and Regional Data Management: Udms Annual 2011: Delft, The
Netherlands. p. 47-58.
http://giscience.uni-hd.de

GIScience
University of Heidelberg

73/12

From IndoorOSM to CityGML (3)


Question:
How can the newly developed IndoorOSM extension be utilized for
the creation of highly detailed CityGML models?
Answer:
It is possible to create highly detailed LoD3 (models with doors,
windows etc.) and LoD4 (models with interior details) by purely using
IndoorOSM data (with respect to some limitations such as
"Manhattan World"-like buildings, roofs etc.)

Goetz M. (2012): Towards generating highly detailed 3D CityGML models from


OpenStreetMap. International Journal of Geographical Information Science (IJGIS). Taylor &
Francis.

02/15/2012

GIScience
University of Heidelberg

Some issues with 3D-VGI

No 3D support in (OSMs) simple data model


Complex 3D modelling is difficult

Not enough tools for users to contribute various 3D


information in different LoDs
Users must be supported/enabled to directly map 3D content

http://giscience.uni-hd.de

GIScience
University of Heidelberg

75

You might also like