Statistical characterization of SAR data: The multiplicative model and extensions

In this paper a comprehensive account of the distributions associated to three SAR image formats (namely, complex, intensity and amplitude) and types of classes (homogeneous, heterogeneous and extremely heterogeneous) is presented, along with their main properties and relations. The backscatter of extremely heterogeneous targets is modelled by the square root of generalized inverse Gaussian distribution, that yields to an extension of K distributions for the return. Modelling with K distributions is applied to SAR data over amazonian areas, showing the discriminatory capability of the methodology. For the sake of completeness, some other distribution that do not fit in the multiplicative model (but that have proved their usefulness) are discussed.