on fourier and wavelets: representation, approximation and ...vetterli/talks/mtns06.pdf · wavelet...
TRANSCRIPT
![Page 1: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/1.jpg)
Audiovisual CommunicationsLaboratory
On Fourier and Wavelets:On Fourier and Wavelets:Representation, Approximation andRepresentation, Approximation and
CompressionCompression
Martin VetterliEPFL & UC Berkeley
Fall 2006
![Page 2: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/2.jpg)
Fall 2006 - 2
AcknowledgementsAcknowledgements
Collaborations: Sponsors: NSF Switzerland• T.Blu, EPFL• M.Do, UIUC• P.L.Dragotti, Imperial College• P.Marziliano, NIT Singapore• I.Maravic, EPFL• R.Shukla, EPFL• C.Weidmann, TRC Vienna
Discussions and Interactions:• A.Cohen, Paris VI• I. Daubechies, Princeton• R.DeVore, Carolina• D. Donoho, Stanford• M.Gastpar, Berkeley• V.Goyal, MIT• J. Kovacevic, CMU• S. Mallat, Polytech. & NYU• M.Unser, EPFL
![Page 3: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/3.jpg)
Fall 2006 - 3
OutlineOutline
1. Introduction through History
2. Fourier and Wavelet Representations
3. Wavelets and Approximation Theory
4. Wavelets and Compression
5. Going to Two Dimensions: Non-Separable Constructions
6. Beyond Shift Invariant Subspaces
7. Conclusions and Outlook
![Page 4: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/4.jpg)
Fall 2006 - 4
OutlineOutline
1. Introduction through History• From Rainbows to Spectras• Signal Representations• Approximations• Compression
2. Fourier and Wavelet Representations3. Wavelets and Approximation Theory4. Wavelets and Compression5. Going to Two Dimensions: Non-Separable Constructions6. Beyond Shift Invariant Subspaces7. Conclusions and Outlook
![Page 5: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/5.jpg)
Fall 2006 - 5
From Rainbows to From Rainbows to SpectrasSpectras
Von Freiberg, 1304: Primary and secondary rainbowNewton and Goethe
![Page 6: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/6.jpg)
Fall 2006 - 6
Signal Representations (1/2)Signal Representations (1/2)
1807: Fourier upsets the French Academy....
Fourier Series: Harmonic series, frequency changes, f0, 2f0, 3f0, ...But... 1898: Gibbs’ paper 1899: Gibbs’ correction
Orthogonality, convergence, complexity
![Page 7: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/7.jpg)
Fall 2006 - 7
Signal Representations (2/2)Signal Representations (2/2)
1910: Alfred Haar discovers the Haar wavelet“dual” to the Fourier construction
Haar series:• Scale changes S0, 2S0, 4S0, 8S0 ...• orthogonality
![Page 8: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/8.jpg)
Fall 2006 - 8
Theorem 1 (Shannon-48, Whittaker-35, Nyquist-28, Gabor-46)Theorem 1 (Shannon-48, Whittaker-35, Nyquist-28, Gabor-46)If a function f(t) contains no frequencies higher than W cps, it is completelydetermined by giving its ordinates at a series of points spaced 1/(2W)seconds apart.[if approx. T long, W wide, 2TW numbers specify the function]
It is a representation theorem:• , is an orthogonal basis for BL• f(t) in BL can be written as
… slow…!
Note:• Shannon-BW, BL sufficient, not necessary.• many variations, non-uniform etc• Kotelnikov-33!
![Page 9: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/9.jpg)
Fall 2006 - 9
Representations, Bases and FramesRepresentations, Bases and FramesIngredients:
• as set of vectors, or “atoms”,• an inner product, e.g.
• a series expansion
Many possibilities:• orthonormal bases (e.g. Fourier series, wavelet series)• biorthogonal bases• overcomplete systems or frames
Note: no transforms, uncountable
![Page 10: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/10.jpg)
Fall 2006 - 10
Approximations, Approximations, aproximationaproximation……
The linear approximation method
Given an orthonormal basis for a space S and a signal
the best linear approximation is given by the projection onto a fixed sub-spaceof size M (independent of f!)
The error (MSE) is thus
Ex: Truncated Fourier series project onto first M vectors corresponding to largest expected inner products, typically LP
![Page 11: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/11.jpg)
Fall 2006 - 11
The The Karhunen-Loeve Karhunen-Loeve Transform: The Linear View (1/2)Transform: The Linear View (1/2)Best Linear Approximation in an MSE sense:
Vector processes., i.i.d.:
Consider linear approximation in a basis
Then:
Karhunen-Loeve transform (KLT):For 0<M<N, the expected squared error is minimized for the basis {gn}where gm are the eigenvectors of Rx ordered in order of decreasingeigenvalues.
Proof: eigenvector argument inductively.Note: Karhunen-47, Loeve-48, Hotelling-33, PCA, KramerM-56, TC
![Page 12: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/12.jpg)
Fall 2006 - 12
Compression: How many bits for Mona Lisa?Compression: How many bits for Mona Lisa?
![Page 13: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/13.jpg)
Fall 2006 - 13
A few numbersA few numbers……
D. Gabor, September 1959 (Editorial IRE)"... the 20 bits per second which, the psychologists assure us, the humaneye is capable of taking in, ...”
Index all pictures ever taken in the history of mankind•
Huffman code Mona Lisa index• A few bits (Lena Y/N?, Mona Lisa…), what about R(D)….
Search the Web!• http://www.google.com, 5-50 billion images online, or 33-36 bits
JPEG• 186K… There is plenty of room at the bottom!• JPEG2000 takes a few less, thanks to wavelets…
Note: 2(256x256x8) possible images (D.Field)
Homework in Cover-Thomas, Kolmogorov, MDL, Occam etc
![Page 14: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/14.jpg)
Fall 2006 - 14
Source Coding: some backgroundSource Coding: some background
Exchanging description complexity for distortion:• rate-distortion theory [Shannon:58, Berger:71]• known in few cases...like i.i.d. Gaussians (but tight: no better way!)
or -6dB/bit
• typically: difficult, simple models, high complexity (e.g. VQ)• high rate results, low rate often unknown
![Page 15: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/15.jpg)
Fall 2006 - 15
New image coding standard New image coding standard …… JPEG 2000 JPEG 2000
Old versus new JPEG: D(R) on log scale
Main points:• improvement by a few dB’s• lot more functionalities (e.g. progressive download on internet)• at high rate ~ -6db per bit: KLT behavior• low rate behavior: much steeper: NL approximation effect?• is this the limit?
![Page 16: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/16.jpg)
Fall 2006 - 16
New image coding standard New image coding standard …… JPEG 2000 JPEG 2000
From the comparison,• JPEG fails above 40:1 compression• JPEG2000 survives
Note: images courtesy of www.dspworx.com
![Page 17: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/17.jpg)
Fall 2006 - 17
Representation, Approximation and Compression: Why does it matter anyway?Representation, Approximation and Compression: Why does it matter anyway?
Parsimonious or sparse representation of information is key in• storage and transmission• indexing, searching, classification, watermarking• denoising, enhancing, resolution change
But: it is also a fundamental question in• information theory• signal/image processing• approximation theory• vision research
Successes of wavelets in image processing:• compression (JPEG2000)• denoising• enhancement• classification
Thesis: Wavelet models play an important role
Antithesis: Wavelets are just another fad!
![Page 18: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/18.jpg)
Fall 2006 - 18
OutlineOutline
1. Introduction through History2. Fourier and Wavelet Representations
• Fourier and Local Fourier Transforms• Wavelet Transforms• Piecewise Smooth Signal Representations
3. Wavelets and Approximation Theory4. Wavelets and Compression5. Going to Two Dimensions: Non-Separable Constructions6. Beyond Shift Invariant Subspaces7. Conclusions and Outlook
![Page 19: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/19.jpg)
Fall 2006 - 19
Fourier and Wavelet Representations: SpacesFourier and Wavelet Representations: Spaces
Norms:
Hilbert spaces:
Inner product:
Orthogonality:
Banach spaces:
CP spaces: p-times diff. with bounded derivatives -> Taylor expansions
Holder/Lipschitz α : locally α smooth (non-integer)
![Page 20: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/20.jpg)
Fall 2006 - 20
ExampleExample
consider and
p < 1: quasi norm, p -> 0: sparsity measure
![Page 21: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/21.jpg)
Fall 2006 - 21
More SpacesMore Spaces
Sobolev Spaces WS(R)
If then f is n-times continuously differentiable
Equivalently decays at
Besov Spaces with respect to a basis (typically wavelets)
or wavelet expansion has finite norm
![Page 22: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/22.jpg)
Fall 2006 - 22
AA Tale of Two Representations: Fourier versus WaveletsTale of Two Representations: Fourier versus Wavelets
Orthonormal Series Expansion
Time-Frequency Analysis and Uncertainty Principle
Then
Not arbitrarily sharp in time and frequency!
![Page 23: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/23.jpg)
Fall 2006 - 23
Local Fourier Basis?Local Fourier Basis?
The Gabor or Short-time Fourier Transform
Time-frequency atoms localized at
When “small enough”
Example: Spectrogram
![Page 24: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/24.jpg)
Fall 2006 - 24
The Bad NewsThe Bad News……
Balian-Low Theorem is a short-time Fourier frame with critical sampling
then either
or: there is no good local orthogonal Fourier basis!
Example of a basis: block based Fourier series
Note: consequence of BL Theorem on OFDM, RIAA
![Page 25: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/25.jpg)
Fall 2006 - 25
The Good News!The Good News!
There exist good local cosine bases.
Replace complex modulation by appropriate cosine modulation
where w(t) is a power complementary window
Result: MP3!Many generalisations…
![Page 26: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/26.jpg)
Fall 2006 - 26
Example of time-frequency tiling, state of the art audio coderExample of time-frequency tiling, state of the art audio coder
In this example, it switches from 1024 channels down to 128, makes for prettycrisp attacks!
It also makes the RIAA nervous….
![Page 27: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/27.jpg)
Fall 2006 - 27
AnotherAnother Good News!Good News!
Replace (shift, modulation)
by (shift, scale)
or
then there exist “good” localized orthonormal bases, or wavelet bases
![Page 28: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/28.jpg)
Fall 2006 - 28
Examples of basesExamples of bases
Haar Daubechies, D2
![Page 29: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/29.jpg)
Fall 2006 - 29
Wavelets and representation of piecewise smooth functionsWavelets and representation of piecewise smooth functions
Goal: efficient representation of signals like
where:• Wavelet act as singularity detectors• Scaling functions catch smooth parts• “Noise” is circularly symmetric
Note: Fourier gets all Gibbs-ed up!
![Page 30: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/30.jpg)
Fall 2006 - 30
Key characteristics of wavelets and scaling functions (1/3)Key characteristics of wavelets and scaling functions (1/3)
Daubechies-88, Wavelets from filter banks, ortho-LP with N zeros at ,
Scaling function:
Orthonormal wavelet family:
Scaling function and approximations• Scaling function spans polynomials up to degree N-1
• Strang-Fix theorem: if has N zeros at multiples of (but theorigin), then spans polynomials up to degree N-1
• Two scale equation:
• Smoothness: follows from N,
![Page 31: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/31.jpg)
Fall 2006 - 31
Key characteristics of wavelets and scaling functions (2/3)Key characteristics of wavelets and scaling functions (2/3)
Lowpass filters and scaling functions reproduce polynomials• Iterate of Daubechies L=4 lowpass filter reproduces linear ramp
Scaling functions catch “trends” in signals
scalingfunction
linearramp
![Page 32: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/32.jpg)
Fall 2006 - 32
Key characteristics of wavelets and scaling functions (3/3)Key characteristics of wavelets and scaling functions (3/3)
Wavelet approximations• wavelet has N zeros moments, kills polynomials up to degree N-1• wavelet of length L=2N-1, or 2N-1 coeffs influenced by singularity at each
scale, wavelet are singularity detectors,• wavelet coefficients of smooth functions decays fast,
e.g. f in cP, m << 0
Note: all this is in 1 dimension only, 2D is another story…
![Page 33: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/33.jpg)
Fall 2006 - 33
How about singularities?How about singularities?
If we have a singularity of order n at the origin(0: Dirac, 1: Heaviside,…), the CWT transform behaves as
In the orthogonal wavelet series: same behavior, but only L=2N-1coefficients influenced at each scale!• E.g. Dirac/Heaviside: behavior as and
Wavelets catch and characterize singularities!
![Page 34: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/34.jpg)
Fall 2006 - 34
Thus: a piecewise smoothThus: a piecewise smooth signal expands as:signal expands as:
• phase changes randomize signs, but not decay• a singularity influence only L wavelets at each scale• wavelet coefficients decay fast
![Page 35: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/35.jpg)
Fall 2006 - 35
OutlineOutline
1. Introduction through History2. Fourier and Wavelet Representations3. Wavelets and Approximation Theory
• Non-linear approximation• Fourier versus wavelet, LA versus NLA
4. Wavelets and Compression5. Going to Two Dimensions: Non-Separable Constructions6. Beyond Shift Invariant Subspaces7. Conclusions and Outlook
![Page 36: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/36.jpg)
Fall 2006 - 36
From linear to non-linear approximation theoryFrom linear to non-linear approximation theory
The non-linear approximation methodGiven an orthonormal basis for a space S and a signal
the best nonlinear approximation is given by the projection onto an adaptedsubspace of size M (dependent on f!)
The error (MSE) is thus
andDifference: take the first M coeffs (linear) or
take the largest M coeffs (non-linear)
set of M largest
![Page 37: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/37.jpg)
Fall 2006 - 37
From linear to non-linear approximation theoryFrom linear to non-linear approximation theory
Nonlinear approximation• This is a simple but nonlinear scheme• Clearly, if is the NL approximation scheme:
This could be called “adaptive subspace fitting”From a compression point of view, you “pay” for the adaptivity
• in general, this will costbits
These cannot be spent on coefficient representation anymore
LA: pick a subspace a priori NLA: pick after seeing the data
![Page 38: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/38.jpg)
Fall 2006 - 38
Non-Linear Approximation ExampleNon-Linear Approximation Example
Nonlinear approximation power depends on basisExample:
Two different bases for• Fourier series• Wavelet series: Haar wavelets
Linear approximation in Fourier or wavelet bases
Nonlinear approximation in a Fourier basis
Nonlinear approximation in a wavelet basis
![Page 39: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/39.jpg)
Fall 2006 - 39
Non-linear Approximation ExampleNon-linear Approximation Example
Fourier basis: N=1024, M=64, linear versus nonlinear
• Nonlinear approximation is not necessarily much better!
D= 2.7
D= 2.4
![Page 40: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/40.jpg)
Fall 2006 - 40
Non-linear Approximation ExampleNon-linear Approximation Example
Wavelet basis: N=1024, M=64, J=6, linear versus nonlinear
• Nonlinear approximation is vastly superior!
D= 3.5
D= 0.01
![Page 41: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/41.jpg)
Fall 2006 - 41
Nonlinear approximation theory and waveletsNonlinear approximation theory and wavelets
Approximation results for piecewise smooth fcts• between discontinuities,
behavior by Sobolev or Besov regularity• k derivatives ⇒ coeffs when• Besov spaces can be defined with wavelets bases. If
then [DeVoreJL92]:
![Page 42: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/42.jpg)
Fall 2006 - 42
Smooth versus piecewise smooth functions:Smooth versus piecewise smooth functions:
It depends on the basis and on the approximation method
s=2, N=2^16, D_3, 6 levels
![Page 43: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/43.jpg)
Fall 2006 - 43
OutlineOutline
1. Introduction through History2. Fourier and Wavelet Representations3. Wavelets and Approximation Theory4. Wavelets and Compression
• A small but instructive example• Piecewise polynomials and D(R)• Piecewise smooth and D(R)• Improved wavelet schemes
5. Going to Two Dimensions: Non-Separable Constructions6. Beyond Shift Invariant Subspaces7. Conclusions and Outlook
![Page 44: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/44.jpg)
Fall 2006 - 44
Wavelets and CompressionWavelets and Compression
Compression is just one bit trickier than approximation…A small but instructive example:Assume
• , signal is of length N, k is U[0, N-1] and is• This is a Gaussian RV at location k
• Note: Rx = l!
Linear approximation:
Non-linear approximation, M > 0:
![Page 45: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/45.jpg)
Fall 2006 - 45
Given budget R for block of size N:Given budget R for block of size N:
1. Linear approximation and KLT: equal distribution of R/N bits
This is the optimal linear approximation and compression!
2. Rate-distortion analysis [Weidmann:99]
High rate cases:• Obvious scheme: pointer + quantizer
• This is the R(D) behavior for R >> Log N• Much better than linear approximation
Low rate case:
• Hamming case solved, inc. multiple spikes:- there is a linear decay at low rates
• L2 case: upper bounds that beat linear approx.
![Page 46: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/46.jpg)
Fall 2006 - 46
Piecewise smooth functions: pieces are Piecewise smooth functions: pieces are Lipschitz-Lipschitz-αα
The following D(R) behavior is reachable [CohenDGO:02]:
There are 2 modes:• corresponding to the Lipschitz- pieces• corresponding to the discontinuities
![Page 47: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/47.jpg)
Fall 2006 - 47
Lipschitz-Lipschitz-αα pieces: Linear Approximation pieces: Linear Approximation
The wavelet transform at scale j decays as (j << 0)
Keep coefficients up to scale J, or choose a stepsize for a quantizer
Therefore, M ~ 2J coefficientsSquared error:
Rate:• Number of coefficients
Thus
Just as good as Fourier , but local!
![Page 48: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/48.jpg)
Fall 2006 - 48
Rate-distortion behavior for piecewise poly. using an oracleRate-distortion behavior for piecewise poly. using an oracle
An oracle decides to optimally code a piecewise polynomial by allocatingbits “where needed”:
Consider the simplest case
Two approximations errors• : quantization of step location• : quantization of amplitude
Rate allocation: versus
Result:
![Page 49: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/49.jpg)
Fall 2006 - 49
Piecewise polynomial, with max degree NPiecewise polynomial, with max degree N
A. Nonlinear approximation with wavelets having zero moments
B. Oracle-based method
Thus• wavelets are a generic but suboptimal scheme• oracle method asymptotically superior but dependent on the model
Conclusion on compression of piecewise smooth functions:D(R) behavior has two modes:
• 1/polynomial decay: cannot be (substantially) improved• exponential mode: can be improved, important at low rates
![Page 50: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/50.jpg)
Fall 2006 - 50
Can we improve wavelet compression?Can we improve wavelet compression?
Key: Remove depencies across scales:• dynamic programming: Viterbi-like algorithm• tree based algorithms: pruning and joining• wavelet footprints: wavelet vector quantization
Theorem [DragottiV:03]:Consider a piecewise smooth signal f(t), where pieces are Lipschitz- .There exists a piecewise polynomial p(t) with pieces of maximum degreesuch that the residualis uniformly Lipschitz- .
This is a generic split into piecewise polynomial and smooth residual
![Page 51: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/51.jpg)
Fall 2006 - 51
Footprint Basis and FramesFootprint Basis and Frames
Suboptimality of wavelets for piecewise polynomials is due toindependent coding of dependent wavelet coefficients
Compression with wavelet footprints
Theorem: [DragottiV:03]Given a bounded piecewise polynomial of deg D with K discontinuities.Then, a footprint based coder achieves
This is a computational effective method to get oracle performanceWhat is more, the generic split “piecewise smooth” into “uniformly smooth +
piecewise polynomial” allows to fix wavelet scenarios, to obtain
This can be used for denoising and superresolution
![Page 52: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/52.jpg)
Fall 2006 - 52
Denoising Denoising (use coherence across scale)(use coherence across scale)
This is a vector thresholdingmethod adapted to waveletsingularities
![Page 53: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/53.jpg)
Fall 2006 - 53
OutlineOutline
1. Introduction through History2. Fourier and Wavelet Representations3. Wavelets and Approximation Theory4. Wavelets and Compression5. Going to Two Dimensions: Non-Separable Constructions
• The need for truly two-dimensional constructions• Tree based methods• Non-separable bases and frames
6. Beyond Shift Invariant Subspaces7. Conclusions and Outlook
![Page 54: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/54.jpg)
Fall 2006 - 54
Going to Two Dimensions: Non-Separable ConstructionsGoing to Two Dimensions: Non-Separable Constructions
Going to two dimensions requires non-separable basesObjects in two dimensions we are interested in
• textures: per pixel• smooth surfaces: per object!
![Page 55: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/55.jpg)
Fall 2006 - 55
Models of the world:Models of the world:
Gauss-Markov Piecewise polynomial the usual suspect
Many proposed models:• mathematical difficulties• one size fits all…• reality check• Lena is not PC, but is she BV?
![Page 56: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/56.jpg)
Fall 2006 - 56
Current approaches to two dimensionsCurrent approaches to two dimensions……..
Mostly separable, direct or tensor products
Fourier and wavelets are both direct product constructions
Wavelets: good for point singularities but what is needed are sparsecoding of edge singularities!• 1D: singularity 0-dimensional (e.g. spike, discontinuity)• 2D: singularity 1-dimensional (e.g. smooth curve)
DWT
![Page 57: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/57.jpg)
Fall 2006 - 57
Recent work on geometric image processingRecent work on geometric image processing
Long history: compression, vision, filter banks
Current affairs:
Signal adapted schemes• Bandelets [LePennec & Mallat]: wavelet expansions centered at
discontinuity as well as along smooth edges• Non-linear tilings [Cohen, Mattei]: adaptive segmentation• Tree structured approaches [Shukla et al, Baraniuk et al]
Bases and Frames• Wedgelets [Donoho]: Basic element is a wedge• Ridgelets [Candes, Donoho]: Basic element is a ridge• Curvelets [Candes, Donoho]
Scaling law: width ~length2
L(R2) set up• Multidirectional pyramids and contourlets [Do et al]
Discrete-space set-up, l(Z2)Tight frame with small redundancyComputational framework
![Page 58: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/58.jpg)
Fall 2006 - 58
Nonseparable Nonseparable schemes and approximationschemes and approximationApproximation properties:
• wavelets good for point singularities• ridgelets good for ridges• curvelets good for curves
Consider c2 boundary between two csts
Rate of approximation, M-term NLA in bases, c2 boundary• Fourier: O(M-1/2)• Wavelets: O(M-1)• Curvelets: O(M-2) Note: adaptive schemes, Bandelets: O(M-α)
![Page 59: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/59.jpg)
Fall 2006 - 59
The prune-join The prune-join quadtree quadtree algorithmalgorithm
• polynomial fit to surface and to boundary on a quadtree• rate-distortion optimal tree pruning and joining
quadtree with R(D) pruning R(D) Joining of “similar” leaves
Note: careful R(D) optimization!
![Page 60: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/60.jpg)
Fall 2006 - 60
Multiresolution Multiresolution directional directional filterbanks filterbanks and and contourlets contourlets [M.Do][M.Do]
Idea: find a direct discrete-space construction that has good approxi-mation properties for smooth functions with smooth boundaries• directional analysis as in a Radon transform• multiresolution as in wavelets and pyramids• computationally easy• bases or low redundancy frame
Background:• curvelets [Candes-Donoho] indicate that “good” fixed bases do exist for
approximation of piecewise smooth 2D functions• a frequency-direction relationship indicates a scaling law• an effective compression algorithm requires
- close to a basis (e.g. tight frame with low redundancy)- discrete-space set up and computationally efficient
Question:• can we go from l(Z2) to L(R2), just like filter banks lead to wavelets?
Answer:• contourlets!
![Page 61: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/61.jpg)
Fall 2006 - 61
Directional Filter Banks [Directional Filter Banks [BambergerSBambergerS:92, :92, DoVDoV:02]:02]
• divide 2-D spectrum into slices with iterated tree-structured f-banks
fan filters
quincunxsampling
shearing
![Page 62: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/62.jpg)
Fall 2006 - 62
Iterated directional filter banks: efficient directional analysisIterated directional filter banks: efficient directional analysis
Example:
![Page 63: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/63.jpg)
Fall 2006 - 63
Example of basis functionsExample of basis functions
• 6 levels of iteration, or 64 channels• elementary filters are Haar filters• orthonormal directional basis• 64 equivalent filters, below the 32 “mostly horizontal” ones are shown
This ressembles a “local Radon transform”, or radonlets!• changes of sign (for orthonormality)• approximate lines (discretizations)
![Page 64: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/64.jpg)
Fall 2006 - 64
Adding Adding multiresolutionmultiresolution: use a pyramid!: use a pyramid!
Result:• “tight” pyramid and orthogonal directional channels => tight frame• low redundancy < 4/3, computationally efficient
![Page 65: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/65.jpg)
Fall 2006 - 65
Basis functions: Wavelets versus Basis functions: Wavelets versus contourletscontourlets
Wavelets Contourlets
![Page 66: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/66.jpg)
Fall 2006 - 66
Pyramidal directional filter bank expansion: ExamplePyramidal directional filter bank expansion: Example
Pepper image and its expansion
Compression, denoising, inverse problems:if it is sparse, it is going to work!
![Page 67: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/67.jpg)
Fall 2006 - 67
Approximation propertiesApproximation properties
Wavelets
Contourlets
![Page 68: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/68.jpg)
Fall 2006 - 68
An approximation theoremAn approximation theorem
Curvelets lead to optimal approximation, what about contourlets?
Result [M.Do:03]Simple B/W image model with c2 boundaryContourlet with scaling w ~ l2 and 1 directional vanishing momentThen the M-term NLA satisfies
Proof (very sketchy...):- Amplitude of contourlets ~ 2-3j/4 and coeffs ~2-3j/4 l3jkn
- Three types of coefficients (significant which match direction insignificant that overlap, and zero)- levels 3J and J, respectively, leading to M ~ 23J/2
- squared error can be shown to be ~ 2-3J
![Page 69: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/69.jpg)
Fall 2006 - 69
Example: Example: denoising denoising with with contourletscontourlets
original noisy
wavelet13.8 dB
countourlets 15.4 dB
![Page 70: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/70.jpg)
Fall 2006 - 70
OutlineOutline
1. Introduction through History2. Fourier and Wavelet Representations3. Wavelets and Approximation Theory4. Wavelets and Compression5. Going to Two Dimensions: Non-Separable Constructions6. Beyond Shift Invariant Subspaces: Finite Rate of Innovation
• Shift-Invariance and Multiresolution Analysis• A Variation on a Theme by Shannon• A Representation Theorem
7. Conclusions and Outlook
![Page 71: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/71.jpg)
Fall 2006 - 71
Shift-Invariance and Shift-Invariance and Multiresolution Multiresolution AnalysisAnalysis
Most sampling results require shift-invariant subspaces
Wavelet constructions rely in addition on scale-invariance
Multiresolution analysis (Mallat, Meyer) gives a powerful framework.Yet it requires a subspace structure.
Example: uniform or B-splines
Question: can sampling be generalized beyond subspaces?
Note: Shannon BW sufficient, not necessary
![Page 72: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/72.jpg)
Fall 2006 - 72
A Variation on a Theme by ShannonA Variation on a Theme by Shannon
Shannon, BL case: or 1/T degrees offreedom per unit of time
But: a single discontinuity, and no more sampling theorem...
Are there other signals with finite number of degrees of freedom per unitof time that allow exact sampling results?=> Finite rate of innovation
Usual setup:
x(t): signal, h(t): sampling kernel, y(t): filtering of x(t) and yn: samples
![Page 73: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/73.jpg)
Fall 2006 - 73
A Toy ExampleA Toy Example
K Diracs on the interval: 2K degrees of freedom. Periodic case:
Key: The Fourier series is a weighted sum of K exponentials
Result: Taking 2k+1 samples from a lowpass version of BW-(2K+1)allows to perfectly recover x(t)
Method: Yule-Walker system, annihilating filter, Vandermonde system
Note: Relation to spectral estimation and ECC (Berlekamp-Massey)
![Page 74: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/74.jpg)
Fall 2006 - 74
A Representation Theorem [VMB:02]A Representation Theorem [VMB:02]
For the class of periodic FRI signals which includes• sequences of Diracs• non-uniform or free knot splines• piecewise polynomials
there exist sampling schemes with a sampling rate of the order of the rateof innovation which allow perfect reconstruction at polynomial complexity
Variations: finite length, 2D, local kernels etc
![Page 75: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/75.jpg)
Fall 2006 - 75
A local algorithm for FRI samplingA local algorithm for FRI sampling
The return of Strang-Fix!
local, polynomial complexity reconstruction, for diracs and piecewisepolynomials
![Page 76: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/76.jpg)
Fall 2006 - 76
ConclusionsConclusions
Wavelets and the French revolution• too early to say?• from smooth to piecewise smooth functions
Sparsity and the Art of Motorcycle Maintenance• sparsity as a key feature with many applications• denoising, inverse problems, compression
LA versus NLA:• approximation rates can be vastly different!
To first order, operational, high rate, D(R)• improvements still possible• low rate analysis difficult
Two-dimensions:• really harder! and none used in JPEG2000...• approximation starts to be understood, compression mostly open
Beyond subspaces:• FRI results on sampling, many open questions!
![Page 77: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/77.jpg)
Fall 2006 - 77
OutlookOutlook
Do we understand image representation/compression better?• high rate, high resolution: there is promise• low rate: room at the bottom?
New images• plenoptic functions (set of all possible images)
• non BL images (FRI?)• manifolds, structure of natural images
Distributed images• interactive approximation/compression• SW, WZ, DKLT...
![Page 78: On Fourier and Wavelets: Representation, Approximation and ...vetterli/talks/MTNS06.pdf · Wavelet approximations • wavelet has N zeros moments, kills polynomials up to degree N-1](https://reader030.vdocuments.site/reader030/viewer/2022040600/5e88e27885ab1524672ff5df/html5/thumbnails/78.jpg)
Fall 2006 - 78
PublicationsPublications
For overviews:• D.Donoho, M.Vetterli, R.DeVore and I.Daubechies, Data
Compression and Harmonic Analysis, IEEE Tr. on IT, Oct.1998.• M. Vetterli, Wavelets, approximation and compression, IEEE
Signal Processing Magazine, Sept. 2001
Coming up:• M.Vetterli, J.Kovacevic and V.Goyal,
The World of Fourier and Wavelets:Theory, Algorithms and Applications,2007… ;)