Statistical Modeling of Wavelet Transformed Sar Images

In this paper we model speckle as signal-dependent additive noise, and propose to use the normal inverse Gaussian distribution as a statistical model for the wavelet coefficients of both the reflectance image and noise image. Further, we show how to estimate the parameters specifying these distributions from the parameters included in the K-distribution modeling the intensity of a SAR image. we show that the proposed statistical models make excellent fit to the distributions of the wavelet coefficients of single-look SAR images.