Factorized Geometrical Autofocus for Synthetic Aperture Radar Processing

This paper describes a factorized geometrical autofocus (FGA) algorithm, specifically suitable for ultrawideband synthetic aperture radar. The strategy is integrated in a fast factorized back-projection chain and relies on varying track parameters step by step to obtain a sharp image; focus measures are provided by an object function (intensity correlation). The FGA algorithm has been successfully applied on synthetic and real (Coherent All RAdio BAnd System II) data sets, i.e., with false track parameters introduced prior to processing, to set up constrained problems involving one geometrical quantity. Resolution (3 dB in azimuth and slant range) and peak-to-sidelobe ratio measurements in FGA images are comparable with reference results (within a few percent and tenths of a decibel), demonstrating the capacity to compensate for residual space variant range cell migration. The FGA algorithm is finally also benchmarked (visually) against the phase gradient algorithm to emphasize the advantage of a geometrical autofocus approach.