A Practical Amplitude Distribution for SAR and POLSAR Images

Abstract:
Probabilistic models abound for the amplitude distribution of SAR images. The light-tailed Rayleigh amplitude distribution arises assuming Gaussian real and imaginary parts in the radar returns. Yet, for the most part, the amplitude returns are known to be heavy-tailed. Several heavy-tailed distributions have been proposed to fit these returns. Some claim that they model the underlying physics and others claim they model the complex nature of the heavy-tailed real and imaginary parts of the distribution. The Generalized Γ Distribution (G Γ D) is explored as a fit to the amplitude because it is practical and intuitive. Reliable estimates for the parameters are presented. The model generalizes naturally to mult-look images.