Least-squares design of 2-D FIR filters that are subject to peak error constraints

Abstract

This paper discusses a least-squares error design of 2-D digital filters under a peak error constraint. Initially we present a filter design procedure in which the peak error constraint is satisfied by iterating the design based on a weighted least-squares error criterion to update the weights. The computational complexity required in the filter design is then reduced by switching from the above procedure to the Lagrange method of unknown constants during the course of iteration. Finally, the trade-off between square error and absolute peak error is verified by using a numerical example. © 1998 Scripta Technica, Electron Comm Jpn Pt 3, 81(9): 75–82, 1998