Some features of non-parametric algorithms application in the tasks of strong correlated 2D fields processing

Numerous works are dedicated to non-parametric methods application in tasks of one-dimensional signal detection. Methods for the multi-dimensional case are considered only in the common setting of the task. With the development of processing an image which is a 2D signal, there is a need for detailed analysis of the behaviour of these methods behavior under new conditions. We consider a rank algorithm used for image brightness steps detection which is inherent to this category of detectors. The rank algorithm supposes processing of two samples: the operating sample and reference or noise sample