Spatial-domain design of two-dimensional recursive digital filters

A technique is proposed to design stable two-dimensional (2-D) recursive digital filters that best approximate, in thel_p-sense, a prescribed impulse or step response, using a local state-space model. The Davidon-Fletcher and Powell optimization algorithm is used to minimize the performance measure. Several designs for first-quadrant and asymmetric half-plane filters are solved to illustrate the proposed technique.