One-dimensional linear shift- variant digital filtering using two-dimensional techniques

Abstract

This correspondence investigates some problems associated with the implementation of one-dimensional (1-D) linear shift-variant (LSV) digital filters. By mapping 1-D input and output sequences into two-dimensional (2-D) sequences, we may characterize the input-output relationship of a 1-D LSV digital filter as a 2-D convolution. In this way, a 1-D LSV digital filter may be approximately implemented as a 2-D recursive difference equation. The advantages and disadvantages of this implementation technique are discussed.