The paper is devoted to an employment of least square error estimator for modeling data in a variety of applications. A two-dimensional empirical signal is considered in case of stationarity in wide sense noise embedding and known variance. An adaptive 2D-discrete Fourier estimator is given in details. Its algorithm performances are evaluated in case of a known signal embedded in noise and the application of the 2D-adaptive Fourier estimator to image processing. The developed algorithm is found to be able to closely model the behavior of the neuronal columnar modules located in the primary visual cortex, as long as these are shown to be able to exploit a spatial analysis of the sensory stimuli.
Reviewer: T.Semerdjiev (Sofia)