Optimal "focusing" for low resolution ScanSAR

Deals with the focusing of low resolution ScanSAR data, for both detected amplitude images and interferometric applications. The SAR reference is exploited to achieve ScanSAR focusing in conventional techniques. Such techniques provide quite effective compensation of the azimuth antenna pattern (e.g. no scalloping) when the azimuth time-bandwidth product of the ScanSAR echo is large, but fail to do so as the burst shortens, being reduced to an ineffective weighting of the output. The result is an azimuth varying distortion of the focused impulse responses, a distortion that is partly compensated for in the multilook average (not available for interferometric applications) at the price of a reduction in the processed Doppler bandwidth. This paper proposes quite a different approach. A set of short kernels, each suitable for "focusing" at a specific azimuth bin, has been optimized to reconstruct source reflectivity in the minimum mean square error sense. That pseudoinversion converges to the "conventional" focusing when the burst extent is large and for short bursts, edge effects are accounted for. These azimuth-varying kernels can be suitably tuned to meet constraints in the resolution/sidelobes trade-off and have proved capable of providing fairly undistorted output and fine resolution. They better exploit the available Doppler bandwidth, maximizing the number of looks and the interferometric quality. A decomposition is suggested that implements the inverse operator as a fast preprocessing to be followed by a conventional ScanSAR processor.