Generating correlated gamma random fields with application to synthesis of simulated SAR imagery

A computationally efficient method is described for simulating high-resolution synthetic aperture radar (SAR) imagery. The method involves generating spatially correlated random numbers statistically distributed according to the gamma product model (K-distribution), a non-Gaussian model that has received a great deal of attention in the literature recently. The spatial correlation is created by exploiting the reproductive property of the gamma distribution, and is computed by summing elements of an uncorrelated, gamma distributed random field with properly chosen parameters. Comparisons are presented between the simulated SAR imagery and real SAR imagery gathered with the MIT Lincoln Laboratory millimeter-wave SAR