Urban-Target Recognition by Means of Repeated Spaceborne SAR Images

Abstract

The relative low resolution (~25 m times 5 m on the ground) of spaceborne C-band synthetic aperture radar (SAR) data as acquired, for example, by European Space Agency sensors ERS and Envisat, can be significantly increased (up to submeter precisions) by processing coherently long series of images. Moreover, by analyzing the amplitude of the radar signal and by exploiting polarization diversity, the main radar characteristics of urban targets can be estimated, and a system for automatic recognition of a set of scattering structures can be developed. In this paper, we study the variation of the amplitude of the received radar signal as a function of the acquisition geometry [normal baseline and Doppler centroid (DC)] to retrieve the extension of the targets in range and azimuth. The dependence of the radar amplitude on temperature at the time of acquisition has been discovered to be very useful to identify extended resonating targets. Dihedrals are discriminated from specular or trihedral reflectors through the phase of Envisat alternating polarization (AP) acquisitions. By means of all gathered radar measurements, the bases for the development of a system for the automatic recognition of six main typologies of urban SAR targets (ground-level and elevated backscatterers, simple and resonating dihedrals, poles and trihedrals) have been laid. Radar data are then combined with in situ surveys and aerial photos, allowing a first assessment of the methodology in urban area.