8. Conclusions

It was shown that considering the 2D Fourier transformation as an overdetermined inverse problem could result in a procedure with increased noise rejection capability. In order to find a robust method the iteratively reweighted least squares procedure using Cauchy-Steiner weights is proposed. In the framework of the new inversion-based FT method series expansion is used for discretization of the complex Fourier spectrum. The procedure is relatively quick, due to the appropriate choice of the set of basis function: the Hermite functions are involved, as they are eigenfunctions of the Fourier transformation.
