Creating a 3D virtual city with CityGML data by LOD1 and LOD2 of Schleswig-Holstein

Master Thesis at ifp - Navid Shamshirbandi

Navid Shamshirbandi

Creating a 3D virtual city with CityGML data by LOD1 and LOD2 of Schleswig-Holstein

Duration of the Thesis: 6 months
Completion: October 2019
Supervisor: Dr.-Ing. Volker Walter
Examiner: Prof. Dr.-Ing. Uwe Sörgel


 

Introduction

The number of 3D city model which are being created has increased since the last decade. The variety of methods for creating the 3D data provided the opportunity to have 3D data. Since 2008 which CityGML data has issued by the Open Geospatial Consortium in the first version, we have the new XML file which includes information about the semantic attributes. So the virtual cities which are created based on this data have the capability for different GIS tasks and also for complex analysis, urban planning, hazard situations, and different simulation tasks. The possibility to extend the data provides the opportunity for different analyses and tasks. CityGML data is used for exchanging, representing and storing the virtual city models and the main focus of the data is on the semantic attributes. Data also structured in five different levels of detail from zero to four in which zero is the footprint and no 3D data and level one is the coarsest visualization of 3D data without any information about roof shape. In level two information about the roof shape and roof type is provided in level three we have more information about the facades and in level four indoor information is included. Every state office for surveying and Geoinformation in Germany has the data in level one and two. Schleswig-Holstein state also possesses the rich amount of 3D data in LOD1 and LOD2 which sells these products to the users and in this work we want to provide the method for storing a large amount of CityGML data into a database and update them and also recommending a suitable web-application for visualization of the data and also in the end enabling the users to download the CityGML data directly from database for further analysis.

Methodology and result

The structure of this thesis is divided into:

  • Storing s large amount of CityGML data into database
  • Update the database regularly
  • Visualization of CityGML data into a web
  • Downloading the CityGML data directly from the database
Figure 1: workflow of master thesis

For storing the data, we used Importer/Exporter which is the software from the 3DCityDatabase package. Regardless of the size of data, we imported all buildings for Schleswig-Holstein in LOD1 and LOD2 into the database. The package provides the schema for storing the geometry and semantic attributes. Importer/Exporter also provides the functionality for extracting the attributes of interest from the database. the other advantage of having the database, it enables us to convert the data into a visualizable format like JSON, COLLADA, and KML. For updating the database there is a package named virtualCITYSYSTEM which provides the VirtualcityDATABASE software for adding, manipulating and also deleting the building but it is not open source. To do this part, we imported the data from 2017 and 2018 into the FME desktop and finding out which one has changed, then deleting the building from the database by python.

For visualization purposes, the conversion of data from a database is necessary. It could be done either by FME software or Importer/Exporter and since we are working based on a tiling strategy which enables us to load the data which are in our view direction, and we want to include it into the HTML code, conversion to the JSON is needed. For visualization, we worked with the Cesium virtual globe which is based on the JavaScript library and developed by Analytic Graphic Inc. (AGI). 3D-Web-Client which is the part of the 3DCityDatabase has extended the functionality of the CesiumJS and also provided the functionality to explicitly linking the semantic attributes with the geometric. Semantic attributes can be extracted from the database either by python and psycopg2 library and also by the Importer/Exporter. Since we want to include it into the HTML code, we use the online spreadsheet and publish the data into the google cloud. Then by querying from the table, visualization of the semantic attributes is possible.

Figure 2: linking semantic attributes with geometric using online spreadsheet

Here in figure 2, the web client can see both semantic attributes and also thematic attributes when selecting the building. The advantageous of storing the data into the google cloud is, we can manipulate or change the attributes when we execute the URL but the problem with the method of visualization is, in a case of updating, when database changes even for one building, all the database must be converted to visualizable format and it is not possible to include or exclude data from original file. Figure 3 shows the visualization of geometry and semantic attributes which in this case, GMLID, Height of building, story above the ground and functionality are extracted from the database and visible.

Figure 3: Semantic value of the building is visualized by selecting a building

The last part of this thesis is to enable users to download the CityGML file directly from the database. for this part, FME desktop has been used for connecting to Database via Web Feature Service and also FME server for streaming the data. For starting the execution of the FME-server-workspace and hence the multiple downloads, a button was added and linked to an on click-function (Download). This function sends the download request, using the runDataDownload-function provided by the FMEserver-library. in order to send the zip file to us we should use fmedatastreaming service which is called by the GET request via url. Figure 4 indicates that how the data downloaded from the database and visualized by the CityGML viewer.it can be used for further analysis for example when traffic analysis for specific region is needed, they can download all buildings and features to analyze with this enriched data.

Figure 4: Representation of the single building with FME

Conclusion

As a conclusion of this work, we distributed the CityGML data into the web application and also storing the data into the open-source PostgreSQL and recommending the method for updating the database regularly. Users can download the data directly from the database while they are seeing the buildings and can extend the CityGML data with desired features and attributes with ADE (Application Domain Extension) from Importer/Exporter for specific reasons.

Future works

  • Providing a CityGML viewer for the web application, then it is not necessary to store the data into a database and convert them.
  • Providing a method for converting the CityGML data directly to another visualizable format without the necessity for storing to database
  • Providing an OGC compliant WMS standard for CityGML data.
  • The method for calculating the story above the ground which is included in the CityGML data is not accurate. They are by default consider the height of each floor something around 3 meters which is different from building to building.
  • Providing the possibility to connect to the different databases at the same time and retrieving the data of interest. In this master thesis which was aimed at generating the 3D city models from 2 different types of data sets, it was not possible to visualize the city in the same port while providing the WFS service.

Ansprechpartner

Dieses Bild zeigt Volker Walter

Volker Walter

Dr.-Ing.

Gruppenleiter Geoinformatik

Zum Seitenanfang