A Statistical Characterization Of Sea-state SAR Images
Yves Delignon, R. Garello, A. Hillion · 2005
Synthetic Aperture Radar (S AR) images are known to be susceptible to speckle effect due to the coherent technique used in their generation. The intensity I(s) of the observed image at a given point s is first modeled through a statistical estimation of its probability density function ( pdf ) . Then a multiplicative model is used where the intensity I(s) is assumed to be the dot product of the sea-state surface reflectivity S(s) by stationary white noise (s peckle) of known probability density function (p df). In both cases we perform hypothesis tests to select the best candidate distribution. As a result, it is shown that, for SEASAT-A satellite images from the Atlantic ocean, Beta distibutions are good candidates as the pdf for sea-state reflectivity. The method is fairly general and can be applied to images of various sea surface states.