Spotlight mode SAR and ISAR are microwave imaging systems that are used to obtain high resolution reflectivity images by processing data obtained from many different perspective views of a target. Usually a chirp signal with a high time-bandwidth product is transmitted and the returned signal from the target is coherently demodulated and then sampled, providing polar samples of the 2-D Fourier transform of the target reflectivity. The most commonly used image reconstruction algorithm then interpolates the polar data to a Cartesian grid and applies a 2-D FFT. An alternative reconstruction algorithm, called convolution-back-projection, first filters and inverse transforms each demodulated return using a 1-D FFT and then back-projects the result. In this paper we propose a simplified back-projection algorithm in which the filtered projections are obtained automatically by choosing the radar waveform to be the impulse response of the desired filter. The necessary filtering is then accomplished through the physical mechanism of the waveform reflecting off the target, which is described by a convolution. A parallel processing architecture is proposed for the back-projection step and its computational and memory requirements are analyzed. Finally, we describe a second potential method for obtaining projections, which assumes a conventional linear FM waveform.
David C. Munson,
"A New Back-Projection Algorithm For Spotlight-Mode SAR And ISAR", Proc. SPIE 1058, High Speed Computing II, (17 May 1989); doi: 10.1117/12.951673; https://doi.org/10.1117/12.951673