Weighted least-squares design technique for two-dimensional finite impulse response digital filters

A method for designing multidimensional finite impulse response digital filters with minimum weighted mean square error is presented. This technique allows for the design of filters with arbitrary desired frequency characteristics and frequency band weighting. The design equations consist of a system of linear equations with closed-form expressions for the matrix and vector elements. The design of one-dimensional multiband filters, two-dimensional circularly symmetric multiband filters and two-dimensional single-frequency-selective bandpass filters is described in detail to provide illustrative examples of the use of this method. Computer implementation of the design algorithm for these filter types is described and sample numerical results are presented.<
>