Rational approximation of three-dimensional digital filters (Q1092001)

In the paper the problem of approximating a 3-D nonrecursive filter (polynomial transfer function) by a 3-D recursive digital filter (rational transfer function) is discussed. The specific feature of the proposed method consists in the fact that the necessary data are the impulse response coefficients and the autocorrelation function of the sequence of these coefficients. It is shown that in this way it is no longer necessary to solve a nonlinear optimization problem. The solution, i.e., the numerator and denominator coefficients of the rational function, is found by solving a linear set of equations. The special case when the denominator is separable is shown to be very useful in applications. Finally, an illustrative example is presented. The paper generalizes to 3-D digital filters similar techniques obtained for 1-D and 2-D digital filters.