Showing papers on "Quadrature mirror filter published in 2004"

On the Choice of Band-Pass Quadrature Filters

Abstract: Band-pass quadrature filters are extensively used in computer vision to estimate information from images such as: phase, energy, frequency and orientation,1 possibly at different scales and utilise this in further processing-tasks. The estimation is intrinsically noisy and depends critically on the choice of the quadrature filters. In this paper, we first study the mathematical properties of the quadrature filter pairs most commonly seen in the literature and then consider some new pairs derived from the classical feature detection literature. In the case of feature detection, we present the first attempt to design a quadrature pair based on filters derived for optimal edge/line detection. A comparison of the filters is presented in terms of feature detection performance, wherever possible, in the sense of Canny and in terms of phase stability. We conclude with remarks on how our analysis can aid in the choice of a filter pair for a given image processing task.

Efficient implementation of nearly perfect reconstruction FIR cosine-modulated filterbanks

Abstract: In this paper, a new fast implementation of nearly perfect reconstruction finite impulse response (FIR) cosine-modulated filterbanks is shown. A new expression for the analysis bank has been developed, and based on it, an efficient implementation structure is obtained for specific prototype filter lengths. The new computationally efficient algorithm can be applied to both the conventional cosine-modulated filterbank and the Koilpillai-Vaidyanathan pseudo-quadrature mirror filterbank. We also show how the implementation cost (measured in multiplications and additions per input sample) is reduced.

Method and system for transmission of information data over a communication line

Abstract: A transceiver system and method comprises a transmitter and a receiver. The transmitter includes a synthesis filter bank and an overlapping unit. The synthesis filter bank generates and modulates a set of wavelets. The overlapping unit generates a multicarrier signal carrying the information data and provides it to the communication line. The receiver includes a separating unit having a separating shift register and an analyzing filter bank having an N-points FFT unit and a demodulator coupled thereto. The method includes at the transmitter end providing a sequence of prototype wavelets by using N-points IFFT, modulating the sequence of the prototype wavelets by said information data, overlapping the modulated prototype wavelets to generate a multicarrier signal carrying the information data. The method includes at the receiver end separating overlapped wavelets from the multicarrier signal, analyzing the separated wavelets by using an N-points FFT, and demodulating spectral frequency amplitudes of the wavelets.

Discrete wavelet transform based on cyclic convolutions

Abstract: A filterbank in which cyclic convolutions are used in place of linear convolutions will be referred to as a cyclic convolution filterbank (CCFB). A dyadic tree-structured CCFB can be used to perform a discrete wavelet transform suitable for coding based on symmetric extension methods. This paper derives two types of efficient implementation techniques for the tree-structured CCFB: one using complex arithmetic and one using only real arithmetic. In addition, the present paper analyzes in detail the perfect reconstruction (PR) condition for the two-channel CCFB and shows that this condition is much less restrictive than that of usual two-channel filterbanks. When each of the two-channel CCFBs constituting a tree-structured CCFB is designed to be PR, the whole system is PR. A quadrature mirror filter (QMF) CCFB having the PR property is demonstrated to be easily designed using a standard filter design subroutine. In contrast, designing a linear-phase PR FIR QMF bank that has good frequency response is not possible when filters are realized by linear convolutions.

Orthogonality detector, and quadrature demodulator and sampling quadrature demodulator using detector thereof

Abstract: To detect phase mismatches between in-phase and quadrature signals of a quadrature demodulator. The phase mismatches can be detected using the signals obtained by removing high frequency components of output of a multiplier by a low pass filter, the output being the product of the in-phase signals of which low frequency components are removed by a first high pass filter by the quadrature signals of which low frequency components are removed by a second high pass filter.

Quadrature detector, quadrature demodulator using the same, and sampling quadrature demodulator

Abstract: PROBLEM TO BE SOLVED: To detect an orthogonal phase error between an in-phase signal and an orthogonal signal of a quadrature demodulator. SOLUTION: A multiplier 122 produces a product of the in-phase signal from which a low-frequency component is eliminated by a first high-pass filter 111 and the orthogonal signal from which the low-frequency component is eliminated by a second high-pass filter 112. The orthogonal phase error can be detected by the signal from which an output high-frequency component is eliminated by a low-pass filter 131. COPYRIGHT: (C)2010,JPO&INPIT

Signal processing architecture for modulation and interpolation

Abstract: A system and method for oversampling and modulating a digital data stream. A programmable quadrature interpolating filter is provided which allows the passband of the output signal to be selected from a set of mode selections. The set of mode selections provide overlapping passbands spanning the Nyquist range of the filter. The filter allows a signal to be shifted to a desired carrier frequency before the signal is oversampled and filtered, reducing power consumption in devices such as digital-to-analog converters.

A family of IIR two-band orthonormal QMF filter banks

Abstract: The design and characteristics of orthonormal two-band QMF filter banks, with perfect reconstruction and linear phase properties, are considered in this paper. The analysis and synthesis filter banks are implemented using all pass filters. Filters in the synthesis bank are ant causal and unstable filters and the block processing technique and an appropriate causal filter are applied for their real time application. The generated filter banks characteristics and the finite block length influence of the block processing technique applied for ant causal filtering are illustrated for a rectangular input signal case. The corresponding wavelet and scaling functions, generated after five iterations of the analysis bank in a low pass branch, are shown.

Design of approximate Hilbert transform pair of wavelets with exact symmetry [filter bank design]

Abstract: This paper presents a new technique for designing pairs of filter banks whose corresponding wavelet functions are approximate Hilbert transforms of each other. The filters have exact linear phase which yields biorthogonal wavelets with exact symmetry. The technique is based on matching the frequency response of a given odd-length filter bank with an even-length filter bank. The class of EBFB (even-length Bernstein filter bank) is utilized in the matching design. The EBFB has perfect reconstruction and vanishing moments properties structurally imposed and this simplifies the design process. The design is achieved through a non-iterative least squares method.

Quadrature mirror filter design without transition specification

Abstract: The design of two channel linear phase quadrature mirror filter (QMF) banks is considered with constraints on the peaks of the reconstruction error and that of the subband filter ripple sizes. We further considered the design specification without explicitly specified transition band for the subband filters. An iterative algorithm is proposed to design QMF banks that can incorporate with various design constraints. We have suggested the possible design tradeoff between transition band bandwidth and the ripple size of the subband subband filter and/or the reconstruction error. Various design examples are presented to illustrate the versatility of the digital QMF banks obtained by the proposed design method. The algorithm is proved to converge and was found to converge efficiently in all the simulations.

A new approach for estimation of wavelets with non-separable kernel from a given image

Abstract: In this paper, we propose a new approach for the estimation of wavelet with non-separable kernel from a given image. The paper discusses method to the design of biorthogonal 2D-FIR perfect reconstruction filter banks where the filters are estimated from a given image. The design method is proposed with different sampling lattices. The important contribution of this paper lies in the fact that these filters are estimated from given image itself. Several design examples are presented to show the efficacy of the approach and also the ease of design. Since these wavelets are estimated from given image, thus they are called 2-D matched wavelets.

A direct design framework for a class of oversampled perfect reconstruction filter banks

Abstract: A novel framework for the design of oversampled perfect reconstruction filter banks in which all analysis and synthesis filters have the length of twice the downsampling factor is presented. It is shown that given a set of analysis filters, using only simple matrix operations, we can obtain the corresponding synthesis filters which yield perfect reconstruction. In the proposed approach, it is not necessary to employ any iterative optimization technique in order to obtain the synthesis filters. The analysis is performed based solely on observations in a linear vector space. Some design examples are provided and open problems are stated.

Implementation of QMF filters bank under several configurations on transputer card B008

Abstract: In this paper, we present the study and the simulation of a quadrature mirror filter (QMF) bank and the effect of the parallelism when it is implemented on a transputer card B008.

Wavelets filter banks based on continuous-time asymptotic filters

Abstract: Digital signal processing has played a key role in development of telecommunication systems over the last decade. In recent years digital filter banks have been occupying an increasingly important role in wireless and wireline communication systems. In some applications it is necessary to use filter banks which are not uniform. These decompose a given spectrum into sub-spectra of different bandwidths. Filter banks with exponentially spaced center frequencies and bandwidths are of particular interest. The best known examples of this type are octave filter banks, which are considered here. Closely related to these are dyadic wavelets, which are transformation kernels used for multiresolution analysis of non-stationary signals, the so-called wavelet analysis. In this paper an obvious and straightforward idea of wavelet transform construction and building of the corresponding digital filter bank based on a class of finite order error signal energy optimal IIR filters called asymptotic filters are presented and illustrated by simulation

Simulation of cosine modulated pseudo QMF banks application to speech

Abstract: In this paper, we present the study and the simulation of a quadrature filter bank based on cosine modulation with application to speech coding.

Design and VLSI implementation of QMF banks

Abstract: The authors consider the design of multirate filterbanks for applications such as subband coding with IIR QMF (quadrature mirror filter) pairs. These offer reduced complexity and low latency at the expense of the loss of exact linear phase. In particular, consideration is given to the use of all-pass sections where linear phase is approximately achieved by being part of the objective in numerical optimisation experiments. This approach compares favourably with previous IIR based approaches. Finite wordlength design using simulated annealing shows that low coefficient wordlength may be used. This leads to efficient realisations with three-port adaptors. Using pipelining implementation, a flexible VLSI architecture is designed that can be used for a variety of subband decompositions. Layout and simulation of the design have been performed.

A factorization approach for symmetric bi-orthogonal double density wavelet filter banks

Abstract: We look at the design of a class of oversampled filter banks and the resulting framelets. The oversampled property is achieved via an extra sub-band resulting in double density filter banks (DDFB's). We design a class of such filters with linear phase property.

Signal extension for orthogonal filter banks and its application to image coding

Abstract: In this paper, a signal extension method is proposed for realization of nonextension convolution in a two-channel orthogonal FIR filter bank. Methods of performing nonextension convolution include the symmetric extension method for a linear phase filter bank and the periodic extension method applicable to an arbitrary filter bank. However, the periodic extension method is not appropriate for image compression because it becomes discontinuous at the signal edge. In this paper, for application to an orthogonal filter bank, the input signal of the decomposed filter bank is extended in such a way that the input signal in the synthesized filter bank has symmetry similar to that in the symmetric extension method. In this way, nonextension convolution is realized. In this case, it is shown that there are degrees of freedom in the extension signals determined by the input signal and the coefficients of the decomposed filter bank. The extended signal is calculated by singularity decomposition in such a way that the extended signal becomes smooth. Finally, the proposed method is applied to the coding of static images and the effectiveness of the method is demonstrated. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 87(3): 41–48, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10125

Optimal design of QMF filter banks based on max-flat polynomial for texture analysis

Abstract: In this paper, we present an optimum design of two-hand Finite Impulse Response (FiR) Quadrature Mirror Filter hank (QMF) based upon max-flat polynomial for maximum possible discrimination between textured images. There are several applications which may not require reconstruction of signal from its transformed coefficients e.g. texture analysis, remote sensing etc. For such applications, features are extracted at different frequency resolution scales. Therefore, it is extremely important that transformed image coefficients should be distortion less. In view of the fact, that in practice, it is impossible to obtain error free coefficients, we present an optimal design method for maximum possible discrimination between any particular class of textures with minimum possible coefficient error. This method employs pass-band response and transition width of a filter as an objective function to design an optimum filter bank.

Design of two-channel perfect reconstruction filter bank with specified LPF magnitude frequency response

Abstract: This paper addresses the problem of designing a perfect reconstruction filter bank with a specification on the magnitude response of the prototype low pass filter. It is shown that adopting a particular ITR form for the squared magnitude frequency response of the low pass filter allows one to generate a PRFB. A method for matching this form to a desired magnitude frequency response is developed. An example is provided to illustrate the approach.

Mirroring Properties of Sub-band Adaptive Beamforming Arrays using Quadrature Mirror Filter Banks

Abstract: The concept of the sub-band adaptive beamforming arrays (ABA) is derived from the conventional fullband adaptive antenna arrays. Instead of doing the beamforming for the entire frequency band as in the full band ABA, only one sub-band of the frequency band is subjected to the same in here. The sub-banding of the full frequency band is done here by a quadrature mirror filter (QMF) bank. In a previous attempt, an approximate method has been proposed to reproduce the higher sub-band output from the lower sub-band output [1]. However, the higher sub-band has higher centre frequency, which demands higher Nyquist bandwidth. This will increase the operational complexity of the sub-band ABA. Recently it has been shown that the QMF sub-band adaptive beamforming can be performed by optimising the higher sub-band output and then, multiplying the signal output with a transforming factor that generates the lower sub-band output [2]. The lower frequency band requires only lower Nyquist bandwidth. It is believed that this will reduce the hardware complexity, the cost and the workload of the earlier version of the QMF sub-band ABA system [1]. In this work, a hardware implementation and its characteristics of such a sub-band ABA to generate the combined output from the higher sub-band is investigated.

Equaripple digital filters in quadrature mirror filter banks for nuclear magnetic tomography

Abstract: The wavelet transform represents a very favourite tool for removing noise 6om signal on many areas of digital signal processing A problem consists in that selected wavelet functions are mostly used that in advance defme the impulse response of quadrature minor filters (QMF) of the type of lowpass and highpass filters in a bank of filters The paper deals with the design of QMF bank using standard design algorithms for digital filters The Remez algorithm is used here as an example of the design of a bank of partial digital filters