Function approximation to SAR image regions
Quanbing Wu · 1998
This article discusses function approximation to region surfaces in synthetic aperture radar (SAR) imagery. The need for function approximation is discussed from a signal estimation viewpoint in regard to noise filtering and edge detection. Assuming a multiplicative noise model and Gaussian statistics, a maximum likelihood (ML) criterion is formed for function approximation. Simulation results are used to compare the ML estimates of parameters with those of three least square (LS) criteria. The former is shown to be superior.