Two dimensional digital filters for subjective image processing
Abstract
This paper presents a design technique for designing approximately circularly symmetric lowpass, highpass, bandpass, high frequency boost and low frequency boost digital filters for subjective image processing applications. An approach is used which parallels the use of the Butterworth, Chebychev or other type of polynomial approximations to obtain one dimensional lowpass digital recursive filters. The other filter designs are then derived from the lowpass filter design. The designed filters are very close to being circularly symmetric for a wide range of critical frequencies. In the design procedure, the squared magnitude characteristic of the desired circularly symmetric filter is chosen in the Laplace Transform domain. The bilinear transformation is then used to map the squared magnitude characteristic into the two dimensional ZWTransform domain. A pseudostate space representation for the corresponding two dimensional ZWTransform is obtained. The eigenvalues with magnitudes less than one are then used as roots of a denominator polynomial with distinct roots to form the ZWTransform of the stable two dimensional digital filter.
 Publication:

13th Asilomar Conference on Circuits, Systems, and Computers
 Pub Date:
 1980
 Bibcode:
 1980ieee.conf..499A
 Keywords:

 Bandpass Filters;
 Digital Filters;
 High Pass Filters;
 Image Processing;
 Low Pass Filters;
 Network Synthesis;
 Chebyshev Approximation;
 Critical Frequencies;
 Design Analysis;
 Error Analysis;
 Laplace Transformation;
 Mathematical Models;
 Polynomials;
 Recursive Functions;
 Spatial Filtering;
 Instrumentation and Photography