“…Based upon this framework, several nonlocal filtering approaches have been reported in the literature. An example of an efficient nonlocal approach is the NLSAR technique, which is widely used in various applications of multi-channel data, including polarimetric, interferometric, and tomographic reconstructions [11,13,[98][99][100][101]. The similarity measure in NLSAR with multi-channel images is computed using the generalized likelihood criterion, and an expansion of the exponential kernel is used to convert the similarity into a weight.…”