On the true multilook intensity distribution in SAR imagery

The speckle intensity after the multilooking process is generally modelled as a gamma distributed random variable. This is definitely true only if the averaging is performed with uncorrelated samples of equal mean radiometry, each one distributed by a negative exponential function. If the samples still follow this negative exponential distribution, but are correlated or have different mean radiometry, the gamma distribution function is only an approximation of the true probability density function (pdf). J.W. Goodman derived the exact distribution resulting from an incoherent summation of correlated samples with unequal mean power. This exact probability density function was not used to describe the multilook speckle intensity up to now. In this communication, the authors show how to fit the true distribution, mainly when the gamma pdf is still used.