An iterative technique for optimally designing a separable 2D FIR filter

Abstract

The optimal design of a separable two-dimensional (2D) finite impulse response (FIR) filter is a nonlinear programming problem. In this paper, an iterative alternating optimization technique for optimally designing the separable 2D FIR filter in the mini-max sense and the least-square sense is proposed. Implementation of the proposed technique only requires a suitable initial solution. The simulation experiment validates the effectiveness of the proposed technique.

Sparse separable 2‐D FIR filter design based on iterative reweighted ℓ1 norm and greedy searching techniques

“…The separable 2-D FIR filter can be grouped into two categories: frequency-domain design and the sequence-domain design [8]. The former design is based on the decomposition of some desired frequency response matrix.…”

“…Problem (7) can be solved by using TR-IGS technique or the iterative technique proposed in Ref. [8].…”

“…The TR-IGS technique is an effective technique to optimize the coefficients of the separable 2-D FIR filters. Implementation of the TR-IGS technique requires an additional critical parameter, which affects the accuracy and efficiency of this technique [8]. Very recently, a new iterative technique for optimally designing the separable 2-D FIR filters was proposed [8].…”

“…Implementation of the TR-IGS technique requires an additional critical parameter, which affects the accuracy and efficiency of this technique [8]. Very recently, a new iterative technique for optimally designing the separable 2-D FIR filters was proposed [8]. It only requires a suitable initial solution in implementation.…”