DESIGN OF 2-DIMENSIONAL DIGITAL-FILTERS IN THE SPATIAL DOMAIN

The design of two-dimensional (2-D) recursive digital filters in the spatial domain is investigated. Two algorithms are proposed for this purpose, namely, a modified version of the Newton-Raphson (MNR) method, and the well-known Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm. Both of these methods are shown to outperform the popular Davidon-Fletcher-Powell (DFP) algorithm in terms of efficiency, when used for filter design. When the two are compared between themselves, it is seen that the MNR is more efficient than the BFGS for most cases except for very low-order filters, where the BFGS performs slightly better.