Simultaneous mean and texture edge detection in SAR clutter

Abstract

The authors consider simultaneous edge detection of synthetic aperture radar (SAR) images in terms of both the local mean value and a texture parameter related to the depth of modulation. They assume that SAR clutter has K-distributed statistics which can be completely characterised by the mean and order parameter of the distribution. Approximate distributions, based on matching a gamma distribution to the K-distributed amplitude or intensity, are derived. Analytic maximum likelihood tests for the presence of an edge are then derived. Two criteria for optimisation are considered: maximising the total probability of detecting an edge within a window; and maximising the accuracy with which the edge position can be determined. The authors consider each test separately and also investigate a two-stage test which approximately optimises both measures. They indicate the effect of different prior knowledge on the ability to detect edges and also demonstrate the limitations imposed by the approximations made.