Analysis of Speckled Imagery with Parametric and Nonparametric Tests

Synthetic aperture radar (SAR) has a pivotal role as a remote imaging method. Obtained by means of coherent illumination, SAR images are contaminated with speckle noise. The statistical modelling of such contamination is well described according the multiplicative model and its implied G distribution. The understanding of SAR imagery and scene element identification is an important objective in the field. In particular, reliable image contrast tools are sought. Aiming the proposition of new tools for evaluating SAR image contrast, we investigated method based on stochastic divergence based . As a consequence, we proposed several divergence measures specifically tailored for G distributed data. We also introduced a nonparametric approach based on the Kolmogorov-Smirnov distance for G data. Statistic tests based on such measures were also devised and assessed. Their performance were quantified according to the resulting test sizes and powers. Using a Monte Carlo simulation approach, a robustness analysis is presented for several degrees of contamination. It was identified that the proposed tests based on the triangular and the arithmetic-geometric measures outperformed the classical Kolmogorov-Smirnov, Kullback-Leibler and Bhattacharyya approaches.