Variable 2-D FIR digital filter design and parallel implementation

Abstract

The frequency response of a variable two-dimensional (2-D) digital filter is specified by several parameters called spectral parameters. The spectral parameters independently control the shapes of the variable frequency response, It is difficult to apply the existing frequency transformation-based methods to design such general variable 2-D filters. This paper proposes a general approach by assuming each filter coefficient as a multidimensional (M-D) polynomial of the spectral parameters, then the optimal coefficients of the M-D polynomials are found in the least squares sense. Also, the paper shows that the resulting variable 2-D filters can be implemented in parallel form, so that parallel signal processing is possible.