Using SAR data as a source for reconstructing the Earth’s three-dimensional surface is a intensively researched topic. Classical methods such as radargrammetry, SAR interferometry, or SAR tomography are established tools in the Earth observation industry, but rely on at least two different images of the same scene, captured from different viewing angles. However, multiple images of the target scene are not always available or cost-intensive to procure. And, for a number of applications, even a rough estimate of the topography would suffice. In this work, we present the first attempt to use a convolutional neural network (CNN) in order to derive heights from single slant range SAR intensity images. With experiments relying on real high-resolution SAR data, we demonstrate the general feasibility and the potential of SAR-based single-image height reconstruction in urban sceneries.
«Using SAR data as a source for reconstructing the Earth’s three-dimensional surface is a intensively researched topic. Classical methods such as radargrammetry, SAR interferometry, or SAR tomography are established tools in the Earth observation industry, but rely on at least two different images of the same scene, captured from different viewing angles. However, multiple images of the target scene are not always available or cost-intensive to procure. And, for a number of applications, even a r...
»