A realization of 3-D adaptive separable-denominator state-space filters suitable for parallel processing

This paper considers the realization of three-dimensional (3-D) adaptive separable-denominator state-space filters suitable for parallel processing. The least mean squares (LMS) algorithm is used to adjust the filter parameters. First, a 3-D local state-space model which is separable in the denominator is realized in a parallel form. Next, the LMS algorithm is employed to adjust the filter parameters in the parallel form. In addition, the computational time and the number of multipliers (adders) required for the realization are estimated. Finally, in a numerical example, the present adaptive filter is applied to the problem of designing a 3-D digital filter in the spatial domain.