Approximated model based on the optimal realization of a 2-D SDDF

In this paper, we describe an improved algorithm for the 2-D separable denominator digital filter (SDDF). It is based on the reduced-dimensional decomposition of 2-D specifications. After finding two 1-D filter coefficients using singular value decomposition (SVD), we synthesize them into a 2-D filter and approximate the filter by the balanced realization. In order to optimally realize the SDDF, we use the equivalent transform relation between minimum coefficient quantization error and minimum roundoff error. In order to estimate the performance of the improved filter a series of experiments have been carried out using a noise degraded image. The filtered image is compared with Gaussian filtering. The result show that it is degraded by about 0.02%, but is easy to design and more efficient for estimating filter coefficients.