3D Development/Capturing

From OpenStreetMap Wiki
Jump to navigation Jump to search

What are possibilities to capture 3D information in our environment?

The basis of all 3D-related things in OSM is the mapping of the 3D information itself. Nowadays, there are numerous possibilities for interested 3D mappers to capture the data, without the necessity of having expensive surveying equipment, e.g. by the use of Smartphones or just your eyes!

This page is a result of the 2nd 3D Workshop Garching and shall provide a collection of possible approaches/sensors/tools for low-cost 3D mapping and what to take care of while doing so. Also, discussions about what has to be done in order to use this for 3D OSM may take place here.

Feel free to expand the page. For personal comments let's use the talk page, please!


Todays smartphones provide various sensors which can be used for 3D mapping.

  • Camera: taking images, see below
  • 3D Camera: some of the latest smartphones are equipped with two cameras with which stereoscopic measurements are possible. How to use this for 3D mapping?
  • GPS: mainly useful for 2D information like exact building outlines at ground floor level, position of the entrance etc. may also be used for height measurement if roof level is accessible
  • Accelereometer/Gyro: measure angles, extremely useful for measuring building heights, roof angles etc. See for example the Smart Measure app for Android phones.
  • Barometer: height measurement if roof level is accessible


Facade Images

Current building block models could be enhanced significantly by attaching textures from the real building to it. Tagging could also be done remotely if it is possible to call up a street-level picture of the building.

3D models from Images

3D models can in general be calculated from 2D image information (see Photogrammetry), however, the quality of the result is strongly depending on the number and quality of input images as well as the post-processing. There exist several software packages offering this process as a service, e.g. osm-bundler, insight3D OpenSource, Autodesk 123D Catch, My3DScanner, 3DTubeMe, ...

The university of Granada demonstrated an UAV 3D imaging system with 3D modeling with open source VisualSFM[1][2] for buildings.[3]

This technique could potentially be used for future 3D model repositories like OpenBuildingModels or Open3DMap. However, it needs to be evaluated thoroughly how suitable this is compared to manual 3D modelling.

Advantage: Generally easy capturing (just take enough photos), no direct measurement necessary

Aerial imagery ("passive" mapping)

Like in 2D mapping, aerial imagery from Bing might be used to gather 3D information.

  • Modelling roof outlines from aerial images is in general possible and sometimes maybe easier than doing it directly "in the field"
  • One has to be careful due to the displacement of those images, this often leads to very inaccurate results (cf. Roof modelling)
  • Deriving building heights from oblique aerial imagery (e.g. Bing's Bird's eye view?)

Laser meters

Allow for direct, non-contact measurement of arbitrary distances with high accuracy. Low-cost laser meters are already available for "only" less than 100 EUR...

Point cloud data

Devices like the Microsoft Kinect or Asus Xtion are able to capture 3D information as point cloud data. While aquiring the data is as simple as walking around the building while recording, point cloud data sets tend to be large, are noisy, require the registration of multiple views into one model and require triangulation of the resulting data set to receive a mesh. So, while the sensors themselves are relatively cheap, the additional steps required might not be worth it. Libraries usefull for dealing with point cloud data are e.g. the PCL (http://pointclouds.org/).

Use your eyes!

You can already map many 3D-related information just with your eyes, e.g.

  • Number of levels of a building
  • Basic roof type, orientation
  • Facade/Roof colours and materials
  • ...

Kites, Quadrocopters etc.

Acquiring up-to-date aerial imagery by using Kites or the like equipped with cameras.

  • Feasibility? Costs/Effort?
  • Legal issues?

See also