Skip to Main Content
In synthetic aperture radar (SAR) one wishes to reconstruct the reflectivity function of a region on the ground from a set of radar measurements taken at several angles. The ground reflectivity is found by interpolating measured samples, which typically lie on a polar grid in frequency space, to an equally spaced rectangular grid in frequency space, then computing an inverse Fourier transform. The classical polar format algorithm (PFA) is often used to perform this interpolation. We describe two other methods for performing the interpolation and imaging efficiently and accurately. The first is the gridding method, which is widely used in the medical imaging community. The second method uses unequally- spaced fast Fourier transforms (USFFTs), a generic tool for arbitrary sampling geometries. We present numerical and computational comparisons of these three methods using both point scattering data and synthetic X-band radar reflectivity predictions of a construction backhoe.