Paper Info
Title
Electromagnetic Modelling for Information Extraction from High Resolution SAR Images of Urban Areas
Abstract
The launch of new satellites hosting High Resolution (HR) Synthetic Aperture Radar (SAR) sensors has certainly been one of the most interesting novelties in the last year in the field of microwave remote sensing (1). SAR images with 1 meter resolution are completely changing our approach of interpretation of any kind of scene. Whether we consider a natural scenario or a built-up area, high resolution deeply affects the relative SAR image and often confuses and mystifies comprehension for user accustomed to standard spaceborne SAR products with resolution of tens of meters. As a matter of fact, users of standard resolution SAR products might think that HR products allow a "better" view of the scene, whereas HR resolution images allow to view "more" details of the observed scene. These considerations are particularly true and becomes transparent for urban areas where the radar return of a resolution cell can be dominated by man-made objects whose dimensions are significantly smaller than the resolution cell; this is not the case for natural landscapes. For this reason, urban areas need sound and tailored electromagnetic models able to well represent the interactions between the radar signal and the urban structures. However, model-based approaches are rarely presented in the literature including a reliable electromagnetic modelling: simple geometric models are in general introduced. These models do not fully explain the SAR image features as far as also layover and shadowing effects change their appearance in HR SAR images of urban areas or man-made objects. Moreover, surfaces roughness and buildings materials are poorly or not considered at all, thus neglecting the important role they play in the SAR image formation. In conclusion, it is then mandatory to provide an electromagnetic model so that even for simplified structures, any feature on HR SAR images can be explained (2, 3). In this paper we introduce original electromagnetic models to investigate how different resolutions can affect the formation of the SAR images. The proposed models apply under a wide range of hypothesis, thus they are not simply based on the phenomenological observation of SAR images, the latter being an approach that could lead to considerations that are only true for the case at issue. The proposed models rely, instead, on the quantitative description of the scattering mechanisms produced by the most common geometries appearing in urban areas. In particular, appropriate and efficient expressions describing the electromagnetic behavior of plane surfaces, dihedral and trihedral configurations are derived, employed
Year
DOI
Venue
2008
10.1109/IGARSS.2008.4778797
IGARSS
Keywords
Field
DocType
data mining,backscatter,geometric model,high resolution,radar imaging,solid modeling,synthetic aperture radar,feature extraction,image analysis,image resolution,surface roughness,information extraction,electromagnetic modeling
Radar,Computer vision,Radar imaging,Computational electromagnetics,Computer science,Synthetic aperture radar,Remote sensing,Feature extraction,Information extraction,Solid modeling,Artificial intelligence,Image resolution
Conference
Citations 
PageRank 
References 
0
0.34
2
Authors
4
Name
Order
Citations
PageRank
Giorgio Franceschetti142054.26
Raffaella Guida26314.12
Antonio Iodice345373.07
Daniele Riccio4781118.99