Contrast in speckled imagery with stochastic distances

Synthetic aperture radar (SAR), ultrasound-B, laser, and sonar imagery are contaminated with speckle noise. The statistical modelling of such contamination is well described by the multiplicative model, which yields the G⁰ distribution. In particular, reliable image contrast measures are sought in order to discriminate targets. To that end, we present statistical methods based on stochastic divergences and on the Kolmogorov-Smirnov distance for G⁰ data. Their performance is quantified according to their test sizes and powers. A robustness analysis is also presented for several degrees of contamination. We show that the proposed tests based on triangular and arithmetic-geometric measures outperform the Kolmogorov-Smirnov distance.