Abstract
This paper reviews various concepts and solutions of time-invariant and time-varying multirate filter banks. It discusses their performance for image and video coding at low bit rates, and their applicability in the mpeg-4 framework. Time-invariant multirate filter banks, and methods of design with different criteria appropriate for signal compression are first presented. Several procedures of quantization, namely scalar and lattice vector quantization, with bit allocation optimized in the rate-distortion sense, are used for the encoding of the subband signals. A technique of rate-constrained lattice vector quantization (rc-lvq), combined with a three components entropy coding, allow, together with distortion psychovi-sual weighting mechanisms to obtain significant visual improvements versus scalar quantization or the zerotree technique. However, time-invariant multirate filter banks, although efficient in terms of compression, are not well suited for content-based functionalities. Content-based features may require the ability to manipulate and thus encode a given region in the scene independently of the neighbouring regions, hence the use of transformations that can be adapted to arbitrary size bounded supports. Also, to increase the compression efficiency, one may want to adapt the transformation to the region characteristics, and thus use transform switching mechanisms, with soft or hard transitions. Three main classes of transformations can address these problems: shape-adaptive block transforms, transforms relying on signal extensions and transforms relying on time-varying multirate filter banks. These various solutions, with their methods of design, are reviewed. Emphasis is put on an extension of the SDF (symmetric delay factorization) technique which opens new perspectives in the design of time-bounded and time-varying filter banks. A region-adapted rate-distortion quantization algorithm has been used in the evaluation of the transformations compression efficiency. The coding results illustrate the interest of these techniques for compression but also for features such as quality scalability applied to selected regions of the image.
Résumé
Cet article présente une synthèse des principaux concepts relatifs aux bancs de filtres invariants et variants par décalage. Il décrit des méthodes de synthèse, l’efficacité des solutions obtenues, en compression d’images, ainsi que leur application potentielle dans le cadre conceptuel des travaux de normalisation du groupe iso/mpeg4. Les bancs de filtres invariants sont tout d’abord considérés. Les méthodes de synthèse, avec différents critères pertinents pour des applications de compression, sont décrites. Ces solutions sont intégrées dans un algorithme de compression pouvant utiliser par ailleurs plusieurs types de quantification, scalaire et vectorielle sur réseaux réguliers de points (ou treillis) avec une allocation de débit optimisée au sens débit-distorsion. En particulier, une technique de quantification vectorielle en treillis contrainte en débit (rc-lvq), associée à un codage entropique à trois composantes, et à une pondération psychovisuelle des distorsions, permet d’obtenir une amélioration significative de la qualité visuelle, par rapport à la quantification scalaire ou à la technique de « zerotree ». Cependant, ces bancs de filtres invariants bien que performants en terme de compression ne sont pas bien adaptés à des traitements basés sur le contenu de la scène. Ces fonctionnalités « basées contenu » peuvent en effet nécessiter des transformations s’adaptant à des régions de supports bornés et de formes arbitraires. De plus, pour accroître les performances en compression, il peut être souhaitable d’adapter la transformation aux caractéristiques de la région, et ainsi de commuter les transformations. Trois classes principales de transformations peuvent répondre à ces deux problèmes : les transformations par blocs de formes adaptatives, les techniques par extension de signal, et les bancs de filtres variants dans le temps. Les principales solutions et méthodes de synthèse sont décrites. L’accent est porté sur une extension de la technique appelée « factorisation avec retard symétrique » (sdf : symmetric delay factorization), qui ouvre de nouvelles perspectives en synthèse de bancs de filtres variants dans le temps. Un algorithme de quantification vectorielle, optimisé au sens débit-distorsion, et adapté à une analyse en région du signal, est utilisé dans l’évaluation des performances de codage. Les résultats illustrent l’intérêt de telles approches, pour le codage à réduction de débit, mais aussi pour fournir différents niveaux de qualité et de débit sur les différentes régions de l’image.
Similar content being viewed by others
References
MPEG4-Requirements-Group. “MPEG-4 Requirements, version 6.”ISO/IEC JTC1/SC29AVG11/N2075, Feb. 1998).
Onno (P.), Guillemot (C). Data-rate constrained lattice vector quantization: a new quantizing algorithm in a rate-distortion sense,IEEE, ICIP. (Washington). (Oct. 1995).
Linde (Y.).Buzo (A.).Gray (R. M). An algorithm for vector quantizer design,IEEE Trans COM. 28. pp. 84–95. (Jan. 1980).
Shapiro (J. M.), Embedded image coding using Zerotrees of wavelets coefficients.IEEE Trans 41 pp. 3445–3462. (Dec. 1993).
Gilge (M.), Engelhardt (T.), Mehan (R.), Coding of arbitrarily-shaped image segments based on a generalized orthogonal transform,EURASIP Journal on signal Processing and image communication, vol.1, pp. 153–180, (1989).
Chen (H. H.), Civanlar (M. R.). Haskell (B. G.), A block transform coder for arbitrarily-shaped image segments,Proceedings of the Int. Workshop on coding for Very Low Bitrate Video, University of Essex, (April 1994).
Sikora (T.), Makai (B.), Low complex shape-adaptive DCT for generic and functional coding of segmented video,EURASIP Signal Processing: Image Communication, (Nov. 1995).
Barnard (H. J.),Image and video coding using a wavelet decomposition. PhD Thesis, Delft University, (1994).
Nayebi (K.), Barnwell (T. P.), Smith (M. J. T.), Analysis-synthesis systems with time-varying filter bank structures, inIEEE, ICASSP, Vol. IV, pp. 617–620, (March 1992).
Nayebi (K.), Barnwell (T. P.), Smith (M. J. T.), Time-domain filter bank analysis: a new design theory,IEEE Trans 40 pp. 1412–1428, (June 1992).
Aase (S. O.), Image subband coding artifacts : analysis and remedies.PhD Thesis, Norwegian Institute of Technology, Trondheim, Norway,(1993).
Sodagar (I.), Analysis and design of time-varying filter banks.PhD Thesis, Georgia Institute of Technology, (1994).
Herley (C.), Wavelets and filter banks.PhD Thesis, Columbia University, (1993).
Herley (C.), Kovacevic (J.), Ramchandran (K.), Vetterli (M.), Time-varying orthonormal tilings of the time-frequency plane, inIEEE, ICASSP, vol. III, pp. 205–208, (1993).
Herley (C.), Vetterli (M.), Orthogonal time-varying filter banks and wavelets, inIEEE, ISCAS, vol. I, pp. 391–394, (1993).
Herley (C.), Vetterli (M.), Orthogonal time-varying filter banks and wavelets,IEEE Trans. SP 42, pp. 2650–2663, (Oct. 1994).
Herley (C.), Boundary filters for finite-length signals and time-varying filter banks,IEEE Trans. CS-II: Analog and digital signal processing,42 pp. 102–114, (Feb. 1995).
Kalker (T.), On optimal boundary and transition filter« in time-varying filter banks, inIEEE ICIP, pp. 625–628 (Lausanne), (Sep. 1996).
Mertins (A.), Time-varying and support réserva filter banks: design of optimal transition and boundary filters via svd, inIEEE,ICASSP, (Detroit), pp. 1316–1319, (May 1995).
Gopinath (R. A.), Factorization approach to unitary time-varying filter bank trees and wavelets,IEEE, ICASSP,3 (Adelaide, Australia), pp. 109–112, (April 1994).
Gopinath (R. A.), Burrus (S.), Factorization approach to unitary time-varying filter bank trees and wavelets,IEEE Trans. SP, 43, pp. 666–680, (Feb. 1995).
Queiroz (R. L.), On lapped transforms,PhD thesis. University of Texas at Arlington, (1994).
Soman (A. K.), Vaidyanathan (P. P), Nguyen (T. Q.), Linear phase paraunitary filter banks: theory, factorizations and designs,IEEE Trans. SP, 40, pp. 3480–3496, (Dec. 1993).
Queiroz (R. L.de),Rao (K. R.), Optimal orthogonal boundary filter banks, inIEEE, ICASSP, vol. III, pp. 1296–1299, (1995).
Quieroz (R. L.de),Nguyen (T.Q.).Ra (K. R.)The Genloty: generalized linear-phase lapped orthogonal transform,IEEE Trans. SP,44. pp. 497–507 (1996).
Rault (P.), Guillemot (C.), Symmetric delay factorization: a generalized theory for paraunitary filter banks, inEUSIPCO, (1996).
Rault (P.), Guillemot (C.), Symmetric delay factorization: a generalised theory for paraunitary filter banks submitted toIEEE Trans. SP, (1997).
Akansu (A. N.), Haddad (R. A.), Multiresolution signal decomposition.Academic Press (1992).
Vaidyanathan (P. P), Multirate systems and filter banks.Prentice Hall, (1993.)
Vetterli (M.), Kovacevic (J.), Wavelets and subband coding.Prentice Hall (1995).
Esteban (D.), Galand (C.), Application of quadrature mirror filters to split-band coding,Int. Conf. on Acoustics, Speech, and Signal Processing, pp. 191–195, (May 1977).
Johnson (J.D.), A filter family designed for use of quadrature mirror filters to split-band coding,in Proc. ICASSP, pp. 291–294 (April 1980).
Vaidyanathan (P. P.), Regalia (P.), Mitra (S. K.), Design of doubly complementary IRR digital filters using a single complex all-pass filter, with multirate applications,IEEE Trans CS, 34, pp. 378–389, (April 1987).
Smith (M. J. T.), Barnwell (T. P.), Exact reconstruction techniques for three-structured subband coders.IEEE Trans ASSP,3, pp. 434–441, (June 1986).
Nguyen (T. Q.), Vaidyanathan (P. P). Two channel perfect reconstruction FIR QMF structures which yield linear phase analysis and synthesis filters,IEEE Tram. ASSP. 35. pp. 676–690 (May 1989).
Vaidyanathan (P. P.), Multirate digital filters, filter banks, polyphase networks and applications : a tutorial.Proc. IEEE. pp. 56–93, (Jan. 1990).
Ramchandran (K.), Vetterli (M.), Herley (C), Wavelets, subband coding, and best bases,Proc. IEEE, pp. 541–560, (April 1996).
Vaidyanathan (P. P.), Hoang (P. Q.), Lattice structures for optimal design and robust implementation of two-channel perfect reconstruction QMF banks,IEEE Trans, on Acoustics, Speech, and Signal Processing,36, pp. 81–94, (Jan. 1988).
Rioul (O.),Ondelettes régulières: applications à la compression d’images. PhD thesis, ENST Paris, (March 1993).
Onno (P.), Guillemot (C.), Tradeoffs in the design of wavelet filters for image compression,Proc of SPIEconf. on Visual Communication and image Processing, (Boston), (Nov. 1993).
Vetterli (M.), Gall (D. L.), Perfect reconstruction fir filter banks: some properties and factorizations,IEEE Trans ASSP, vol. 37, pp. 1017–1071, (July 1989).
Woods (J. W), O’neil (S. D.), Subband coding of images,IEEE Tram. ASSP 34, pp. 1278–1288, (Oct. 1986).
Vaidyanathan (P. P), Theory and design of M-channel maximally decimated quadrature mirror filters with arbitrary M, having the perfect reconstruction property,IEEE Trans ASSP 35, pp. 476–492, (Apr. 1987).
Smith (M. J. T.), Barnwell (T. P.), A unifyingframework for analysis/synthesis systems based on maximally decimated filter banks, inIEEE ICASSP, pp. 521–524 (March 1985).
Malvar (H.), Signal processing with lapped transforms.Artech House, (1991).
Malvar (H.), Extended lapped transform: fast algorithms and applications,ICASSP, pp. 1797–1800, (May 1991).
Koilpillai (R. D.), Vaidyanathan (P. P), Cosine-modulated fir filter banks satisfying perfect reconstruction,IEEE Trans SP,40, pp. 770–783, (April 1992).
Mau (J.), Perfect reconstruction modulated filter banks,Proc. IEEE ICASSP, pp. IV. 273–IV. 276, (March 1992).
Gopinath (R. A.), Burrus (C. S.), Theory of modulated filter banks and modulated wavelet tight frames,Applied and Computational Harmonic Analysis: Wavelets and Signal Processing, (Aug. 1992).
Nguyen (T. Q.), Koilpillai (R. D.), The design of arbitrary-length cosine-modulated filter banks and wavelets satisfying perfect reconstruction, inProc. IEEE Int. Symposium on Time Frequency and Time Scale Analysis, pp. 299–302, (Oct. 1992).
Nguyen (T. Q.), Koilpillai (R. D.), The theory and design of arbitrary length cosine modulated filter banks and wavelets, satisfying perfect reconstruction,IEEE Trans. on Signal Processing, pp. 473–483, (March 1996).
Guillemot (C.), Onno (P.), Rault (P.), Arbitrary length cosine modulated filter banks : new results and optimization for image compression, submitted toIEEE Trans SP, (1997).
Ramchandran (K.), Vetterli (M.), Best wavelet packet bases in a rate-distortion sense, IEEETrans IP 2, pp. 160–175, (April 1993).
Daubechies (I.), Orthogonal bases of compactly supported wavelets,Communication on Pure and Applied Mathematics, vol XLI, pp. 909–996 (1988).
Conway (J. H.), Sloane (N. J. A.), Fast quantizing and decoding algorithms for lattice quantizers and codes, IEEETrans. IT,28, (March 1982).
Barlaud (M.), Sole (P.), Gadon (T.), Antonini (M.), Mathieu (P.), Pyramidal lattice vector quantization for multiscale image coding,IEEE Trans. IP,3, pp. 367–381, (July 1994).
Vandendokpe (L.), Human visual weighted quantization,Annales télécommunications,47, no. 7-8, pp. 282–292, (1992).
Macq (B.), Weighted optimum bit allocations to orthogonal transforms for picture coding,IEEE Trans. SAC,10, pp. 875–883, (June 1992).
Jensen (E.). Rijkse (K.), Lagendijk (I.) Beek (P. van), Coding of arbitrary-shaped image segments, inProceedings WIASIC’94, p. E2, (1994).
Nuri (V.), A theory for exactly reconstructing size-limited maximally decimated filter banks,Master’s thesis. School of Electrical Engineering and Computer Science, Washington State University, (1993).
Nuri (V.), Bamberger (R. H.). A theory of size limited filter banks,IEEE, ICASSP,3, pp. 161–164, (1993).
Bamberger (R. H.), Eddins (S. L.), Nuri (V), Generalized symmetric extension for size limited multirate filter banks,Tech. rep. School of Electrical Engineering and Computer Science, Washington State University, (1993).
Barnad (H.), Weber (I. H.), Biemond (J.), A region-based discrete wavelet transform, inEUSIPCO, pp. 1234–1237, (Sept. 1994).
Arrowood (J. L.), Smith (M. J. T.), Exact reconstruction analysis/synthesis filter banks with time-varying filters, inIEEE, ICASSP, vol. III, pp. 233–236, (April 1993).
Malvar (H. S.), Signal processing with lapped transforms.Artech House, Inc, (1992).
Bouman (C. A.), Shapiro (M.), A multiscale random field model for Bayesian image segmentation,IEEE Trans IP,3, pp. 162–177, (March 1994).
Author information
Authors and Affiliations
Additional information
Ce travail a été effectué au cnet-ccett. 4 rue du Clos Courtel, F. 35512 Cesson Sévigné, France.
Rights and permissions
About this article
Cite this article
Guillemot, C., Rault, P. & Onno, P. Time-invariant and time-varying multirate filter banks : application to image coding. Ann. Télécommun. 53, 192–218 (1998). https://doi.org/10.1007/BF02997677
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF02997677
Key words
- Image coding
- Standardization
- Audiovisual service
- Digital filter
- Multichannel circuit
- Low rate
- Block quantization
- Invariance
- Time variation
- Wavelet expansion
- Lattice coding