SAR Shape from shading in suburban areas

Abstract

Shape-from-shading (SfS) techniques can be applied to synthetic aperture radar (SAR) images in order to obtain estimates of the slopes and, then, the elevation of the observed scene. Recently the authors proposed a new SfS technique for natural scenarios, based on the use of fractal surface models: problems can arise if non-natural features are present in the observed scene. In this paper, we evaluate the impact of manmade structures on the technique's performance: in particular, the SfS algorithm is applied on a SAR image in which a suburban environment is present. A deep analysis of the main issues is performed and possible solutions are presented and discussed.