Application of the singular-value decomposition for the design of 2-D digital filters

The singular-value decomposition (SVD) design method for design of two-dimensional (2-D) digital filters is surveyed, and several important developments that have led to the present status of the method are described. The SVD of a nonnegative matrix and its application are considered. The SVD of a complex matrix and its application are described. Two variations of the SVD approach which lead to some improvements in the design are described. The method can be applied in conjunction with 1-D design techniques for the design of causal finite impulse response (FIR) or infinite impulse response (IIR), 2-D digital filters with arbitrary amplitude and phase responses. Furthermore, it is relatively simple to apply and leads to a parallel arrangement of pairs of cascaded 1-D filter sections