2-D generalized MFIR filters

Abstract

An efficient structure for 2-D linear phase FIR (finite-impulse response) filters is introduced. The structure is the 2-D counterpart of the 1-D MFIR (multiplicative FIR) filters. It is shown that this structure is capable of satisfying tight specifications with a significantly reduced number of coefficients. The performance improves for tighter specifications with centered transition bands. The filter coefficients are obtained by minimizing an l/sub p/ error function. It is shown that the optimization problem is readily broken up into a set of simpler subproblems. For the special case of p=2, these subproblems reduce to simple matrix inversions. Increasing the filter order does not affect the problem complexity since it only requires an increase in the number of matrix inversions needed. Examples illustrating the capabilities of this structure and comparisons with other approaches are given. >