Skip to Main content Skip to Navigation

Reconnaissance de formes dans des images de télédétection du milieu urbain

Abstract : The Ph.D. studies deal with a new method of the extraction, semi-automatic and hierarchic, of quadrangular urban road network from high spatial resolution imagery. The method is based on a model of streets, and makes use of both the multiresolution analysis and the wavelet transform. The model of streets integrates the geometrical, radiometric and topographical properties of the different classes of streets. It is explicit and generic. A model of road network has been elaborated too. This model is based on the properties of simplified connectivity and hierarchy. The multiresolution analysis, by way of a multiscale analysis of images, allows the extraction of both sides of streets in a class-by-class way. The wavelet transform enables the modeling of information at different characteristic scales, and permits the drawing of their topography by the extraction of their reservations. These extractions have been made by two iterative and multiresolution algorithms. Several quantitative criteria have been defined in cooperation with urban cartographers. They are based on the surface and the location of streets, and on some connectivity indices of the road network. The method has been applied to spaceborne and airborne images of urban areas with different spatial and spectral resolution. The extracted streets are located to around 2 pixels from references whatever the resolution. The method allows a partial automation of the cartographic tasks of urban areas.
Complete list of metadata
Contributor : Lucien Wald Connect in order to contact the contributor
Submitted on : Monday, February 17, 2014 - 4:16:49 PM
Last modification on : Wednesday, November 17, 2021 - 12:30:14 PM
Long-term archiving on: : Sunday, April 9, 2017 - 1:01:19 PM


  • HAL Id : pastel-00948036, version 1


Isabelle Couloigner. Reconnaissance de formes dans des images de télédétection du milieu urbain. Traitement du signal et de l'image [eess.SP]. Université de Nice Sophia-Antipolis, 1998. Français. ⟨pastel-00948036⟩



Record views


Files downloads