Comments on “PolMERLIN: Self-Supervised Polarimetric Complex SAR Image Despeckling With Masked Networks”

A self-supervised despeckling approach based on the decomposition of single-look complex (SLC) synthetic aperture radar (SAR) images into their real and imaginary components has recently been introduced under the name MERLIN. At its core is the observation that, under Goodman’s fully developed speckle model, the real and imaginary parts of single-channel SAR images are independent and identically distributed (i.i.d.). This letter commented here, PolMERLIN, proposes an extension to multichannel SAR images, such as polarimetric SAR (PolSAR) images. This extension is based on the independence between real and imaginary parts of PolSAR images, but this independence generally does not hold, as shown here.