A class of Gaussian-shaped CNN filter banks

In this paper we propose a class of 2D circularly-symmetric filter banks based on a 1D IIR Gaussian prototype. We approach two cases, namely dyadic and uniform filter banks. The Gaussian prototype is obtained using an efficient rational approximation, namely the Chebyshev-Pade approximation. Then, using a frequency transformation, the corresponding 2D filters with circular symmetry are derived. These filters can be implemented both as digital filters, or in the analog version, using for instance the concept of cellular neural networks, which is detailed in the paper.