SAR clutter suppression using probability density skewness

A clutter suppression algorithm is described which exploits asymmetry in the probability density function (pdf) of ground clutter in synthetic-aperture radar (SAR) imagery. In the log-magnitude domain, the probability densities of returns from natural ground clutter and man-made objects exhibit opposite skewness. Using Pearson´s second coefficient of skewness, we have observed negative skewness values for natural clutter regions and positive skewness values for buildings and simulated ground vehicles. Clutter suppression is achieved by attenuating regions having negative skewness relative to regions having positive skewness. Using this approach we have obtained a 9 dB increase in target-to-clutter ratios in SAR images of rural scenes with buildings and other scenes with simulated ground vehicles embedded at various locations. We have also applied the skewness metric to the problem of segmenting man-made objects from clutter backgrounds. Here a slightly modified version of Pearson´s second coefficient of skewness is used as a test statistic analogous to that of an ordered statistic (OS) constant false alarm rate (CFAR) algorithm. Segmentation of simulated ground vehicles embedded in natural ground clutter is demonstrated using the skewness test statistic