SAR interferometric quick-look

Interferometric SAR applications (i.e., DEM generation, differential interferometry) demand high computing time, due to algorithms' complexity and the large amount of data involved. This paper presents an algorithm to compute an interferometric image (fringes) at a reduced ground range resolution (quick look) with a low computational complexity. This result is achieved by means of a raw data “presumming” (i.e., filtering and subsampling) stage, implemented both on range and azimuth directions. The presummed data can then be efficiently compressed by means of monodimensional, small FFT kernels. The algorithm exploits the two surveys' spectral shift by implementing range presumming as a tunable bandpass filtering and subsampling. The altimetric resolution achieved by this technique is the maximum compatible with the range bandwidth and the baseline constraints. The radiometric resolution on the output interferogram is improved with respect to the full resolution product by coherent averaging of the interferograms got by different looks. An efficient integer implementation is presented for the presumming stage, using polyphase filters and lookup table for filter's coefficients. The computing time for a 100×100 km quick-look interferogram is less than 10 minutes for a common Unix workstation