Science Service System

Summary of Proposal MTH1182

TitleAlgorithm development for cardinal effects correction on SAR images.
Investigator Codjia, Claude - Universite du Quebec a Montreal, Geographie
Team Member
Professeur associé Desjardins, Robert - UQAM, Département de géographie
Professeur Cavayas, Françcois - Université de Montréal, Geographie
SummaryThe project’s pr oposed objectives One of the factors limiting the capacity of interpretation of radar images in an urban environment is the variability of the backscattering for similar objects seen under different angles. Indeed, manifest variations of a same image lends to confusion and makes the task more difficult for the interpreter and for the algorithms of classification. The purpose of the present study is to reduce these variations by the perfecting of an algorithm which corrects the image according to the orientation of the urban objects in relation to the plan of illumination of the radar in question. This algorithm of radiometric compensation could allow RSO imagery to reach its full mapping potential in urban environments. This constitutes a useful alternative for the cities, for which optical imagery does not always offer solutions under a quasi-permanent cloud cover. Potential impact for future applications One of the restrictive factors as regards the capacity of interpretation of radar images in urban environments is the variability of backscattering of similar objects under different angles. These variations on the same image bring confusion and renders the task more difficult for the interpreters and for the classification algorithms. The objective of the present study is to reduce the variation between the echoes of the radar through the development of an algorithm which corrects the image according to the orientation of urban objects in relation to the plane of illumination of the radar. This algorithm will facilitate the characterization and the discrimination of the types of ground occupation from RADARSAT-2 images. This algorithm of radiometric compensation could allow the RSO imagery to reach its full mapping capacity in urban environments. This can provide a useful alternative for cities, knowing that optical imagery does not always offer solutions because of the quasi-permanent cloud cover. There can be no doubt that the coming of RADARSAT-2 shall reinforce the use of RSO imagery in urban environment. Method to be used The main purpose of the present work is to research the relationship between the orientation of objects and the backscattering of the radar signal. These relationships shall be established by an in-depth analysis of real images in different modes and in different polarizations through the study of backscattering of identical buildings seen under different angles in relation to the satellite. Simulations from digital models of the city’s surface shall be used to fix the backscattering area of the buildings. The backscattering correction algorithm shall be developped from a backscattering compensation rule. Sectors with an optimal orientation shall be used as reference. Backscattering for other sectors shall be increased according to the azimuthal angle with the satellite’s trajectory. This methodology has already given good results with RADARSAT-1 images. Data requirement: 25 images over 1 year from May 2011(archive + new acquisitions) to May 2012. Reports would be delivered in two periods. The preliminary at the end of autumn 2012 and the final at the end of summer 2013. Depending of the results, scientific papers would be produced.

Back to list of proposals

© DLR 2004-2016