A polarimetric generalized likelihood ratio detector for scattering centers in K-distributed clutter

The problem of detecting targets in a forest using a wideband, polarimetric synthetic aperture radar (SAR) is considered. It is shown that SAR forest measurements are well-modeled by the K-distribution. Histograms of the Huynen parameters of SAR data are displayed in order to describe clutter behavior. Optimal techniques for exploiting polarimetric information to detect point targets are then considered. It is assumed that the point target has an unknown amplitude, unknown orientation about the radar line-of-site, and unknown absolute phase. This assumption leads to a composite-hypothesis detection problem with nuisance parameters. A generalized likelihood ratio test is developed to accommodate the unknowns, and a bounded two-dimensional search for finding their maximum-likelihood estimates is demonstrated.<
>