Approximation analysis of variance errors based on optimal realization of the 2-D SDDF
Taehong Ahn, Jong-An Park · 2002
The paper proposes optimal 2-D separable denominator digital filters (2-D SDDF). The technique is based on decomposing design specifications of 2-D impulse response into two 1-D forms. As coefficient quantization error must be minimized, 2-D SDDF can be optimized after coefficients of balanced realization transformed by an equivalent transformation matrix. The filtering of the noise image is carried out by a space mask from the proposed algorithm and Gaussian filter.