Abstract
In this chapter, expansions of signals into orthogonal or biorthogonal functions, as well as into frames, are formulated as generalized transforms, where the expansion coefficients provide a discrete representation of a signal (globally seen as a generalized transform). The second topic is that of filter banks and subband decomposition, which implement generalized transforms and signal expansions with multirate components, such as decimators and interpolators. It will be shown that such an implementation is possible for generalized transforms that satisfy the condition of periodic shift invariance.
All these topics are preliminary to multiresolution analysis and wavelets, which will be developed in the next chapter. In fact, wavelets may be viewed both as generalized transforms and as signal expansions, and their practical implementation is obtained by filter banks.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Most authors prefer to reserve the term “operator” to the case L 2(I)↦L 2(I), and in the general case they use the term “transformation” (see [9]).
- 2.
The terms “transform” and “transformation” may be confused. We use “transformation” thinking of a model of a system, and “transform” as a mathematical operation, usually accomplished by the inverse operation (inverse transform). “Transform” is also used to denote the result of the application of a transform.
- 3.
Caution must be paid on this term, because in this book “dual” is used throughout for quantities related to the Fourier transform, while here it has a different meaning.
- 4.
The terms “orthogonal” and “orthonormal” will often used interchangeably, unless we want to stress the normalization and then we use “orthonormal.” The same applies to the terms “biorthogonal” and “biorthonormal.”
- 5.
- 6.
Some authors, e.g., Vaidyanathan [12], use these terms to denote down-sampling and up-sampling, respectively.
- 7.
In the figure and in the forthcoming theory we slightly modify the symbolism. The indexes b∈B are replaced by the natural n∈ℤ N to facilitate the matrix notation. The subband components S b (u) are denoted by c n (t) leaving the uppercase to the FT.
- 8.
- 9.
- 10.
In the formulation of tree filter bank we label differently the basic two-channel filter banks at the different stages, whereas other textbooks do not make any distinction, since a single prototype of basic filter bank, e.g., with the same FIR coefficients, is commonly used. We remark, however, that at different stages the filters act with different rates, and they must be regarded as different objects.
- 11.
We recall from Sect. 7.5 that this relation implies that the signal to be processed is modeled as a periodic signal, so that its S/P conversion leads to a constant vector signal.
- 12.
With an abuse of notation, for simplicity, we denote by the same symbol the Fourier transform and the z-transform, e.g., X(f) and X(z) or X(z 0). Also, we will not indicate the convergence region of each z-transform.
References
P. Cassereau, A new class of orthogonal transforms for image processing, Dept. EECS, Mass. Inst. Technol., Cambridge, May 1985
A. Croisier, D. Esteban, C. Galand, Perfect channel splitting by use of interpolators, decimators/tree decomposition techniques, in Int. Symp. on Info., Circuits and Systems, Patras, Greece (1976)
I. Daubechies, Ten Lectures on Wavelets (SIAM, Philadelphia, 1992)
L.E. Franks, Signal Theory (Prentice Hall, Englewood Cliffs, 1969)
R.A. Horn, C.R. Johnson, Matrix Analysis (Cambridge University Press, London, 1985)
N.S. Jayant, P. Noll, Digital Coding of Waveforms (Prentice Hall, Englewood Cliffs, 1984)
J. Kovac̆ević, A. Chebira, An introduction to frames. Found. Trends Signal Process. 2(1), 1–94 (2008)
S. Mallat, A theory of multiresolution signal decomposition: the wavelet representation. IEEE Trans. Pattern Anal. Mach. Intell. 11, 674–693 (1989)
S. Roman, Advance Linear Algebra (Springer, New York, 1992)
M.J. T Smith, T.P. Barnwell, A procedure for designing exact reconstruction filter banks for tree-structured subband coders. Proc. Int. Conf. Acoust. Speech Signal Process. 2, 27.1.1–27.1.4 (1984)
D. Taubman, M.W. Marcellin, JPEG2000, Image Compression Fundamentals and Practice (Kluver Academic, Boston, 2002)
P.P. Vaidyanathan, Multirate Systems and Filter Banks (Prentice Hall, Englewood Cliffs, 1993)
P.P. Vaidyanathan, S.K. Mitra, Polyphase network, block digital filtering, LPTV systems, and alias–free QMF banks: a unified approach based on pseudo circulants. IEEE Trans. Acoust. Speech Signal Process. ASSP–36, 81–94 (1988)
M. Vetterli, J. Kovac̆ević, Wavelets and Subband Coding (Prentice Hall, Englewood Cliffs, 1995)
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A: Proof of Theorem 14.1 on Projections
We first prove that the operator \(\mathcal{P}_{b}\), defined by the kernel (14.57), is idempotent, that is, \(\mathcal{P}_{b}^{2}=\mathcal{P}_{b}\). The kernel of \(\mathcal{P}_{b}^{2}\) is given by
where in {⋅} we have used condition (14.50b) with b=b′. Hence, \(\mathcal{P}_{b}\) is a projector.
The sum of the kernels gives, after use of (14.21),
which states the second of (14.58).
When the self-reciprocal condition holds, it is immediate to check that \(h_{b}^{*}(t_{0}',t_{0})\!=h_{b}(t_{0},t_{0}')\), which states that the projectors \(\mathcal{P}_{b}\) are Hermitian.
Appendix B: Alias-Free Condition in Subband Decomposition
In Sect. 7.7 we have developed the polyphase architecture of a filter, given by an I→J S/P converter, a polyphase filter on J, and a P/S converter (Fig. 14.37). The M×M polyphase matrix p π (t)=[p πji (t)] is obtained from the impulse response p(t 0), t 0∈I, of the filter as
where we suppose that the PD is obtained with the same generator \(A= \{\tau_{0},\ldots,\tau_{M-1}\}\) both at the S/P side and at the P/S side. We see from (14.164) that the polyphase matrix is redundant, since its j,i element depends only on the difference τ j −τ i .
Now, we relate the polyphase matrix to the PD of the impulse response p(t 0), given by
To this end, note that the differences τ j −τ i are points of I and therefore have the unique decomposition
where \(\tau_{\alpha_{ji}}\) is a convenient point of the generator A. Hence, the j,i element of the polyphase matrix can be written in the form
where \(p_{\alpha_{ji}}(t+t_{ji})\) is the polyphase component of p(t 0) of index α ji shifted by −t ji . In conclusion:
Proposition 14.6
For a filter with impulse response p(t 0), t 0∈I, the polyphase matrix p π (t), t∈J, obtained with a cell A=[I/P) of cardinality M, has M 2 elements defined by (14.164). Let p (k)(t)=p(t+τ k ), t∈J,τ k ∈A, be the PD of the impulse response p(t 0), t 0∈I. All the M 2 elements of p π (t) can be obtained from the M elements of the PD p (k)(t) according to (14.167).
As an example, consider a cell A={0,τ 1,τ 2}⊂I of cardinality M=3. Then, the polyphase matrix is
The PD of the impulse response p(t 0) has the 3 components p 0(t)=p(t), p 1(t)=p(t+τ 1), p 2(t)=p(t+τ 2), and we can express the 9 elements of p π (t) in terms of the 3 components p k (t),k=0,1,2. In fact, p(t)=p 0(t), p(t+τ 1)=p 1(t), p(t+τ 2)=p 2(t). For the other 6 elements, we note that −τ 1, −τ 2, τ 2−τ 1, −τ 2+τ 1 can be written in the form \(t_{ij}+\tau_{\alpha_{ij}}\), then, e.g., \(\tau_{2}-\tau_{1}=t_{21}+\tau_{\alpha_{21}}\), and then \(p(t+\tau_{2}-\tau_{1})=p^{(\alpha_{21}}(t+t_{21})\). To be more specific, we continue the example in the 1D case, where A={0,T 0,2T 0}. Then −τ 1=−T 0=−3T 0+2T 0, so that p(t−τ 1)=p (2)(t−3T 0), τ 2−τ 1=T 0, so that p(t+τ 2−τ 1)=p (1)(t), etc. In conclusion,
The symmetry of this matrix is referred to as pseudo-circulant. Evidently, p π (t) is completely determined by the elements of its 0th row, which represent the PD of the impulse response of the filter.
In the general multidimensional case, the symmetry of the polyphase matrix is stated by (14.165) and (14.166), and we referred to it as generalized pseudo-circulant. Proposition 14.6 represents a generalization of a theorem by Vaidyanathan and Mitra [13].
The conclusion is the following:
Proposition 14.7
The alias-free condition in a subband decomposition filter bank requires that the global polyphase matrix p π (t)=(1/M)g π ∗q π (t) must be generalized pseudo-circulant. In this case the global Analysis–Synthesis is equivalent to a filter d(t 0) whose polyphase components are given by the 0th row of p π (t).
Hence, the filter d(t 0) is obtained as the P/S conversion of the M elements of the 0th row of the matrix p π (t) as (Fig. 14.38)
and in the frequency domain, using (7.32)
Once eliminated the aliasing according to the above statement, the perfect reconstruction imposes that d(t 0)=δ I (t 0), which implies that p k (t)=d(t+τ k )=δ I (t+τ k ). In words: the first row of p π (t) must contain pure shifts.
2.1 B.1 Alias-Free Condition in 1D Case
The alias-free condition requires the pseudo-circulant symmetry of global polyphase matrix
This symmetry can be formulated more specifically in the 1D case, where decomposition (14.166) becomes
For instance, for L=4, this gives
With this symmetry, the global subband system becomes a filter with frequency response (see (14.170))
To eliminate the distortion, the further condition is D(f)=1. A solution may be obtained with \(P_{k}(f)=z_{0}^{k}\).
Appendix C: z-Domain Analysis of 1D Subband Decomposition
The frequency-domain analysis developed in Sect. 14.13 can be transferred to the z-domain using the rules outlined in Chap. 11 for the fundamental discrete-time components, where, for the complex variable, we use symbols z for low-rate signals with spacing T and z 0 for high-rate signals with spacing T 0. Since (in this chapter) T=MT 0, the variables are related as \(z=z^{M}_{0}\). We also recall that we find it convenient to define the z-transform as a Haar integral over ℤ(T) or ℤ(T 0) and with the explicit presence of the time spacing, namely
for low-rate and a high-rate signals, respectively. The frequency domain analysis is obtained as a particularization byFootnote 12
At this point we invite the reader to reconsider in detail the second part of Chap. 11 from Sect. 11.5 on, in particular, the rules concerning the passage from the frequency domain to the z-domain. Here, we simply recall the relations in a ℤ(T 0)→ℤ(T) decimator, with T=MT 0, given by
and the relations in a ℤ(T)→ℤ(T 0) interpolator, given by
So, caution must be paid with filters and interpolators, whose frequency relation is the same (but the difference comes from the context).
We also recall the relations of the P/S and S/P conversions, obtained with the cells (14.154). They are respectively (see (11.59) and (11.60))
Using the above rules, we are now ready to obtain the relation in the z-domain from the Fourier analysis. The relationships in the Analysis/Synthesis scheme, given by (14.156), become (considering that the Analysis consists of a decimator and the Synthesis of an interpolator)
By combination we find
and, more explicitly,
The first line represents a filtered version of the original signal, and the rest represents aliasing. Hence, the perfect recovery condition is stated by
i.e.,
3.1 C.1 Polyphase Analysis
The general frequency domain relation (14.129) can be directly rewritten in the z-domain, since they are concerned with ordinary low-rate filters (hence we use the variable z). Thus,
where S(z) is the PD of S(z 0), and Q π (z), G π (z) are the polyphase matrices. Combination of (14.176) gives the global relation \(\tilde{\mathbf{S}}(z)=\mathbf{P}_{\pi}(z)\mathbf{S}(z)\), where
is the global polyphase matrix. Hence, the orthogonality condition becomes
It remains to relate the polyphase vector S(z), with elements S (j)(z), to the z-transform S(z 0) of the input signal s(t 0). The extraction of the polyphase components (S/P conversion) in the z-domain is given by the second of (14.174). Then
and, in matrix form,
Analogously, we can relate the elements of the polyphase matrices Q π (z) and G π (z) to the original subband filters Q(z 0) and G(z 0). Following (14.157), we find
As results from the above relationships, the approach in the z-domain may be viewed as a duplicate of the one carried out in the frequency domain (in the 1D case). Now, the z-domain approach can continue with the investigation on the alias-free and distortion-free conditions. But, now it is a simple exercise to transfer results from the frequency domain to the z-domain.
Rights and permissions
Copyright information
© 2011 Springer-Verlag London Limited
About this chapter
Cite this chapter
Cariolaro, G. (2011). Signal Expansions, Filter Banks, and Subband Decomposition. In: Unified Signal Theory. Springer, London. https://doi.org/10.1007/978-0-85729-464-7_14
Download citation
DOI: https://doi.org/10.1007/978-0-85729-464-7_14
Publisher Name: Springer, London
Print ISBN: 978-0-85729-463-0
Online ISBN: 978-0-85729-464-7
eBook Packages: EngineeringEngineering (R0)