Single‐look light‐burden superresolution differential SAR tomography

Abstract

Research and application is spreading of techniques of coherent combination of complex-valued synthetic aperture radar (SAR) data to extract rich information even on complex observed scenes, fully exploiting existing SAR data archives, and new satellites. Among such techniques, SAR tomography stems from multibaseline interferometry to achieve full-3D imaging through elevation beamforming (spatial spectral estimation). The Tomo concept has been integrated with the mature differential interferometry, producing the new differential tomography (Diff-Tomo) processing mode, that allows ‘opening’ the SAR cells in complex non-stationary scenes, resolving multiple heights and slow deformation velocities of layover scatterers. Consequently, the operational capability limit of differential interferometry to the single scatterer case is overcome. Diff-Tomo processing is cast in a 2D baseline-time spectral analysis framework, with sparse sampling. The use of adaptive 2D spectral estimation has demonstrated to allow joint baseline-time processing with reduced sidelobes and enhanced height–velocity resolution at low computational burden. However, this method requires coherent multilooking processing, thus does not produce full range-azimuth resolution products, as it would be desirable for urban applications. A new single-look adaptive Diff-Tomo processor is presented and tested with satellite data, allowing full range-azimuth resolution together with height–velocity sidelobe reduction and superresolution capabilities and the low computational burden.