Indoor 3D Modeling

3D Models created by a mobile robot can provide a fast and very informative overview of a scene. Most existing approaches use panoramic cameras to acquire the texture information. Unfortunately, the resolution of the resulting images is often greatly decreased by the panoramic mirror, i.e. the wide field of view of the camera. Therefore our approach uses a standard camera, with a narrow field of view compared to panoramic cameras.

The method we use consists of three basic steps:

  1. A single, horizontal laser scannner is used to create a 2D map of the environment, consisting of a set of laser scan points.
  2. Line segments are extracted from the laser map via the Hough Transform. These line segments are further processed, e.g. their length is adapted to ensure that segments representing the corner of a room end in a common point. The resulting line map serves as a floor plan for the 3D model.
  3. Every line segment of the line map represents one wall. The textures for the walls are created from the camera images and are used to provide a photo-realistic view of the scene.

Below you can see pictures of the individual steps and some screenshots of final 3D models created at the WSI.

References

  • [1] Christian Weiss and Andreas Zell, Automatic Generation of Indoor VR-Models by a Mobile Robot with a Laser Range Finder and a Color Camera, in Automone Mobile Systeme (AMS), Stuttgart, Germany, 2005, pp. 107-113

Contact

Christian Weiss, Tel.: (07071) 29-77176, c.weiss at uni-tuebingen.de