Object Reconstruction from Image Sequences

  • Leberl, Franz (Teilnehmer (Co-Investigator))
  • Klaus, Andreas, (Teilnehmer (Co-Investigator))

Projekt: Foschungsprojekt



Matching of image pairs and image sequences is a very important task in the computer vision community. Therefore a lot of effort has already been invested towards increasing accuracy, robustness and performance. The reasons why there is still a lot of research to do in this area, are on the one hand different requirements and on the other hand different intentions. Thus the aim of this work is to develop a new approach appropriate for our main purpose, the reconstruction of facades from calibrated and oriented image sequences. Our requirements are a dense and accurate reconstruction as well as high performance. The latter is essential for our goal to enable the reconstruction of whole city districts from high resolution images within only a few days. In order to allow a simultaneous matching process the accurate orientation of the image sequence is needed. Therefore it is another important goal to analyze existing orientation methods and to develop a suitable fully automatic approach especially for façade reconstruction. The upgrade from the relative to the geo-referenced orientation requires human interaction. By exploiting vanishing points and points of interest this interaction can be minimized, while maintaining high accuracy. The reconstruction of roofs is another task that is addressed in this work. Here the aim is to permit fully automatic reconstruction of building roofs by using data from multiple sensors. not assigned GG: Internal Project of the Institute
Tatsächlicher Beginn/ -es Ende1/06/0114/05/20