Comments on "2-D FIR filters design using least square error with scaling-free McClellan transformation"

Abstract

For original paper see H.C.Lu and K.H.Yeh, ibid., vol.47, pp.1104-07 (2000). The idea of formulating the least-squares error along the desired contour in a quadratic form in the design of two-dimensional (2-D) zero-phase nonrecursive filters has been presented previously (S.S.Kidambi, 1995). Explicit formulas for the design of 2-D fan filters with prescribed inclination have also been presented. As claimed in the original paper, it has been shown that the error depends upon the value of the cut-off frequency of the prototype one-dimensional (1-D) filter and is lowest for a particular value of the 1-D filter cutoff frequency. In fact, the values of t(1, 1) and t(0, 1) presented is, for all practical purposes, the same as those presented by Kidambi. This method has also been extended to the design of circularly-symmetric filters, which the authors of the original paper have done as well. The application of the least-squares method in the design of elliptic filters is fairly straightforward, as has been indicated in the conclusion of the paper by Kidambi.