Reconstruction of buildings from interferometric sar data of built-up areas

U. Soergel, U. Thoennessen, U. Stilla

Research output: Contribution to journalConference articlepeer-review

18 Scopus citations

Abstract

The improved ground resolution of state of the art synthetic aperture radar SAR sensors suggests to utilise such data for the analysis of urban areas. Approaches for a 3D reconstruction of buildings from SAR or InSAR showed good results for rural areas or large and flat buildings, like industrial halls. The situation is more difficult in inner city areas with high buildings which are often located in close proximity to each other. The inherent oblique scene illumination by SAR results in occlusion of objects by more elevated objects interrupting the line of sight. Another problem is the mixture of signal contributions of different objects with the same distance to the sensor. Furthermore, specular reflection and multi-bounce scattering can result in very strong signals which superimpose the backscatter of large parts of their neighbourhood. Geometric constraints of the impact of the mentioned SAR phenomena on the visibility of buildings are derived. The opportunities and limitations of remote sensing of inner city scenes with radar are discussed by comparison of real InSAR data with ground truth like maps and high resolution DEM. An approach for the detection and reconstruction of buildings from InSAR data is proposed and demonstrated.

Original languageEnglish
Pages (from-to)59-64
Number of pages6
JournalInternational Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
Volume34
StatePublished - 2003
Externally publishedYes
Event2003 ISPRS Workshop on Photogrammetric Image Analysis, PIA 2003 - Munich, Germany
Duration: 17 Sep 200319 Sep 2003

Keywords

  • Building reconstruction
  • Interferometric SAR
  • SAR

Fingerprint

Dive into the research topics of 'Reconstruction of buildings from interferometric sar data of built-up areas'. Together they form a unique fingerprint.

Cite this