Recently, the first deep learning-based approaches for the reconstruction of urban height maps from single synthetic aperture radar (SAR) intensity images have been presented. They forgo the need for pairs or stacks of images usually needed for terrain reconstruction with classic multi-image methods such as SAR interferometry, SAR tomography, or radargrammetry. With this paper, we want to identify and discuss those application areas for which single-image height reconstruction can provide a significant benefit over the traditional, well-controlled approaches. We show that single-image-derived height maps are not only relevant in standard 3D reconstruction tasks, but can also support rapid mapping tasks, be used for change detection, provide prior knowledge for processing steps such as phase unwrapping, or aid the visual interpretation of SAR imagery.    «

Recently, the first deep learning-based approaches for the reconstruction of urban height maps from single synthetic aperture radar (SAR) intensity images have been presented. They forgo the need for pairs or stacks of images usually needed for terrain reconstruction with classic multi-image methods such as SAR interferometry, SAR tomography, or radargrammetry. With this paper, we want to identify and discuss those application areas for which single-image height reconstruction can provide a sign...    »