Analysis of Variance under Log-Symmetric Family for SAR Images

In this paper, we relate a family of highly skewed, positive, and flexible random variables to the multiplicative model for SAR images. We relate the parameters of the family of distributions to the backscatter of the terrain and the multiplicative error to the speckle noise. The results show that using a log-symmetric distribution with a heavy tail can be preferred over assuming a Gaussian distribution to ANOVA methodology and a Gamma distribution to speckle noise in SAR images.