Sequential Multidimensional Spectral Estimation

By considering an empirical approximation, and a new class of operators that we will call walking operators, we construct, for any positive ND-toeplitz matrix, an infinite in all dimensions matrix, for which the inverse approximates the original matrix in its finite part. A recursive hierarchical algorithm is presented for sequential dimension spectral representation. A positive comparison in calculus cost, and numerical simulation, for 2D and 3D signals, is also presented.