scientific article; zbMATH DE number 1193385

From MaRDI portal
Publication:3841852





This big paper focuses on the theory and implementation details of hexagonal sampling systems (HSS) (each sample location has exactly six nearest neighbors) and hexagonal quadrature mirror filters (HQMF). The HSS are of particular interest because they exhibit the tightest packing of all regular two-dimensional sampling systems and for a circularly band-limited waveform HSS requires 13.4\% fewer samples than rectangular sampling. This paper is organized in two main sections. The first section presents the theoretical foundation of HSS and HQMF and reviews the theory of HQMF. The rest of the section is organized as follows. Consider a general formulation of an HSS. The HSS is the hexagonal sampling lattice (HSL) \(\lambda\) represented as the set of all vectors \(\overline t= (t_1,t_2)^T\) generated by \[ \overline t= V\overline n= \begin{pmatrix} v_{11} & v_{12}\\ v_{21} & v_{22}\end{pmatrix} \begin{pmatrix} n_1\\ n_2\end{pmatrix}\quad (\text{det }V\neq 0), \] where \(\overline n\) is an integer-valued vector and \(V\) is \(2\times 2\) sampling matrix. The up-sampling and down-sampling in HSS are represented. The process of up-sampling maps a signal on \(\lambda\) to a new signal that is nonzero only at points on the sampling sublattice \(\lambda_k= \{\overline m:\overline m= k\overline n\}\) where \(k\) is \(2\times 2\) nonsingular matrix with integer-valued entries. The output of an up-sampler is related to the input by \[ y(\overline n)= \begin{cases} x(k^{-1}\overline n), &\text{if}\\ 0, &\text{otherwise}.\end{cases} \] The process of down-sampling maps points on the sublattice \(\lambda_k\) to \(\lambda\) according to \(y(\overline n)= x(k\overline n)\) and discards all other points. The redundant analysis/synthesis filter banks (HQMF) in HSS and wavelets that can be obtained by iterating such filter banks are described. Finally, section covers the formulation of the discrete Fourier transform in HSS. In the second section the implementation details of hexagonal multiresolution representation (HMR) are discussed. The rest of the section is organized as follows. Consider a selection of image support for efficient signal processing with hexagonal systems and a filtering in HSS using a hexagonal fast Fourier transform (HFFT) strategy. Systematic application of the HFFT to image processing further restricts sequences to rectangular regions. Processing a rectangularly sampled image (RSI) with an HSS requires the RSI to be mapped into an HSL. The strategy that maps an RSI confined to a square region into a hexagonally sampled image confined to a rectangular region for which it is possible to compute its HFFT is proposed. Finally, this section describes the computation of over-complete HMR (OHMR) by filtering an image with the equivalent filters. An examples of OHMR applied to a digitized radiograph of the breast and of the chest are represented. For the entire collection see [Zbl 0889.00018].











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3841852)