[Home]
[Back Issues]
[Search]
[Vol 1]
[Vol 2]
[Vol 3]
[Vol 4]
[Vol 5]
The Wavelet Digest - Volume 3, Issue 7 (April 27, 1994)
Today's Editor: Wim Sweldens
sweldens@math.sc.edu
Today's Topics:
1. Meeting: UCLA Extension Short Course
2. Question: Mailing list for xwave1-2 software
3. Question: Derivatives of Daubechies wavelets
4. Meeting: Eighth International Conference On Approximation Theory
5. Contents: ACHA, Vol. 1, No. 2
6. Contents: Adv. Comput. Math., Volume 2, No. 3, 1994
7. Question: Page 258 of Recent. Adv. in Wavelet Anal. is missing
8. Question: Looking for references on wavelets on a sphere.
9. Question: Exact reconstruction with Daubechies filters ?
10. Book: Shape in Picture
11. Preprint: Quadratures Involving Polynomials and Daubechies' Wavelets
12. Course: Wavelets for Engineering Applications
13. Question: Looking for a particular function
14. Question: Wavelets in Speech Recognition
15. Contents: JAT Vol. 76 No.3, March. 94
16. Question: Non-Separable Multi-Dimensional Wavelets
17. Answer: WD 3.3 (#9): Multidimensional Factorization
18. Question: Looking for real-time application of wavelets
Submissions for Wavelet Digest:
E-mail to wavelet@math.sc.edu with "submit" as subject.
Subscriptions for Wavelet Digest:
E-mail to wavelet@math.sc.edu with "subscribe" as subject.
To unsubscribe, e-mail with "unsubscribe" followed by your e-mail
address as subject. To change address, unsubscribe and resubscribe.
Archive site, preprints, references and back issues:
Anonymous ftp to ftp.math.sc.edu (129.252.12.3),
directory /pub/wavelet.
Gopher server: gopher.math.sc.edu.
Current number of subscribers: 3347
Notes from the editor:
1. An updated calendar of events will be included in the first issue of
a month. In the meanwhile, it can be obtained from our ftp-site
as the file /pub/wavelet/calendar.
2. Mail sent to the Wavelet Digest between April 16 and 21 might have
gone lost. If your submission does not appear here, please resend
and accept our apologies for this inconvenience.
From: Watanabe, Nonie <NWatanab@UNEX.UCLA.EDU>
Subject: Meeting: UCLA Extension Short Course
UCLA Extension
Department of Engineering, Information Systems and Technical Management
announces a new short course
NONLINEAR DYNAMICS INFORMATION PROCESSING IN CHAOS, FUZZY LOGIC, AND
NEURAL NETWORKS
May 9-11, 1994 at UCLA
Researchers believe they have recently discovered the "Rosetta stone" of
chaos, fuzzy logic, and neural net research. This new artificial neural net
learning methodology can achieve learnable fuzzy logic reasoning at a
specific precision level, an understanding that becomes particularly powerful
for applications in communications and/or control when the inputs are
constrained by real-time nonlinear dynamical data. These new developments and
potential applications have led to the formation of Office of Naval
Research's multi-million dollar five-year core program to study nonlinear
dynamics spatiotemporal information processing (to begin FY '95).
This course discusses application of nonlinear dynamics such as collective
chaos for spatiotemporal information processing that has been embedded
through artificial neural networks (ANN), unifying fuzzy logic with ANN
through the bifurcation cascade of those chaotic outputs generated from
"piecewise negative logic neurons." These chaotic outputs define the fuzzy
membership triangle-shape function with a different degree of precision.
Another advantage of this methodology is the solving of nonlinear dynamics
for information processing in a synthetic nonlinear dynamical environment.
For example, ocean waves can be efficiently analyzed by nonlinear soliton
dynamics, rather than traditional Fourier series.
The course covers the elementary mathematics of nonlinear dynamics, the
bifurcation cascade route to chaos, rudimentary fuzzy logic, and essential
artificial neural network theory for those interdisciplinary participants
with only basic knowledge of the subject area. Various nonlinear dynamical
phenomena in chaos, fuzzy, and neural-like learning are illustrated in terms
of spatiotemporal information processing:
- signal/image de-noise
- control device/machine chaos
- communication coding
- biomedical applications.
Implementation techniques in massive and parallel artificial neural network
chips are given. The course also delineates the difference between the
classical sigmoidal function and the appropriate nonlinear element function
that generates chaos cascade outputs for the fuzzy membership function.
COORDINATOR AND LECTURER
Harold Szu, PhD
Research Physicist, Washington, D.C. Dr. Szu's current research involves
artificial neural networks, chips, fuzzy logic, chaos theory, wavelet
transforms, character recognition, and constrained optimization. Dr. Szu is
also a technical representative to ONR on neural networks and related
research, and has been engaged in plasma physics and optical engineering
research for the past 16 years. He holds six patents, has published more than
150 technical papers, plus two textbooks. Dr. Szu is a co-founder and
president of The International Neural Network Society, and an editor for the
journal Neural Networks.
Dates: May 9-11 (Monday through Wednesday)
Time: 8 am-5 pm
Location: Room 211, UCLA Extension Building, 10995 Le Conte Avenue (adjacent
to the UCLA campus), Los Angeles, California
Fee: $1095, includes lecture notes
For further information, call the Short Course Program Office at (310)
825-3344; or E-mail Marcus Hennessy at mhenness@unex.ucla.edu
From: "Charles D. Kincaid" <statman@apeiron.ufstat.ufl.edu>
Subject: Question: Mailing list for xwave1-2 software
Hello,
Is there interest in a mailing list for the xwave1 and xwave2 software
that mallat, etal. have written? I am using this and extending it for my own
purposes. Every once in a while I have questions about it and there really
isn't a source for answers. I program pretty well (IMHO) and thus can sift
through the code. However, I know that not everyone wants to do that. Also,
even after sifting through the code and determining _what_ it does, there is
sometimes still the question of _why_ it does it.
If you believe that a mailing list for questions regarding xwave1 and
xwave2 would be useful, please send me e-mail at statman@stat.ufl.edu. If
there is suitable interest (a level that I haven't defined yet) I'll start one
up.
Sincerely,
Charles D. Kincaid
Dept. of Statistics Univ. of Florida
statman@stat.ufl.edu (904) 392-1941x206
From: Alistair Rowe <arowe@tartarus.uwa.edu.au>
Subject: Question: Derivatives of Daubechies wavelets
Dear Waveletters,
Dr Paul Abbott and myself have been looking into the so called 'dilational'
derivatives of wavelets such as the Daubechies 4 and Daubechies 6 wavelets.
In truth, the derivatives are of the scaling functions for these cases.
I refer you to Strang Vol 31 No 4 pp614-627, 1989 SAIM Review. He discusses
the connection between the eigenvalues of a system of variables that
correspond to the scaling function values at points within domain of the
scaling function, with the dilational derivatives.
In particular, he says that the first dilational derivative corresponds to
an eigenvalue of 0.5, the second with 0.25 and so on.......
In solving the eigensystem (say for Daubechies 6) we found eigenvalues of
1 (the function), 0.5 (the first derivative), 0.25 (the second derivative),
and another of approximate value -0.27 - What does this correspond to?
Perhaps some sort of fractional derivative?
Alistair Rowe.
From: Charles Chui, Texas A&M University.
Subject: Meeting: Eighth International Conference On Approximation Theory
Eighth International Conference
On Approximation Theory
College Station, Texas
January 8--12, 1995
This conference is dedicated to E. Ward Cheney
in honor of his 65th Birthday
Location: Hilton Conference Center, 801 University Dr., College Station, Texas
Duration: 9:00 a.m., Sunday (Jan. 8) through 4:00 p.m. Thursday (Jan. 12)
Pre-registration: Return the form by October 1, 1994
Registration: 7:00-10:00 p.m., Saturday (Jan. 7), Hilton Hotel Lobby
Plenary Speakers: One-hour invited lectures will be presented by
E. Ward Cheney, Austin, Texas
Charles A. Micchelli, IBM, Yorktown Heights, New York
Wolfgang Dahmen, Aachen, Germany
Gilles Pisier, College Station, Texas, and Paris, France
Ingrid Daubechies, AT&T Bell Laboratories & Princeton
Robert Schaback, Gottingen, Germany
Rong-Qing Jia, Edmonton, Canada
Grace Wahba, Madison, Wisconsin
Doron Lubinsky, Witwatersrand, S. Africa
Popov Prize Winner (to be announced)
The Vasil A. Popov Prize in Approximation Theory: This prize
has been established in memory of Vasil A. Popov and his contributions to
this field. Nominations should be made to Ronald A. DeVore
(devore@math.sc.edu) before June 1, 1994. The winner of this prize
will be asked to deliver a plenary lecture.
Research Talks: All participants are invited to present a
20-minute research talk. These will be organized into parallel sessions.
This year there will be special sessions in several areas, including
splines, wavelets, CAGD, and orthogonal polynomials.
Abstracts: If you wish to present a research talk, please
submit a title and abstract by October 1, 1994. The abstract should be typed
in TeX if at all possible, and should not exceed one page. Please indicate on
your pre-registration form which area your talk fits into, and which days you
would not be able to speak. Electronic copies of the Pre-Registration Form and
the Abstract Form can be obtained via ftp to wavelet1.math.tamu.edu (USER:
texas8 and PASSWORD: texas8) under the file name: texas8.forms.
Proceedings: The proceedings (in two volumes) will
be published by World Scientific Publishing Co. Volume 1 emphasizes
the theoretical aspects and Volume 2, the applied and computational
aspects, of Approximation Theory. Contributed papers are limited to 8 pages.
All papers should be prepared in camera-ready form (using \TeX\ and the
provided macros).
Typing instructions and the macro file will be sent upon receipt of your
abstract. The publisher will provide 25 copies of reprints to be shared by the
coauthors (if any) of each paper. Proceeding volumes will be provided by
paying a slightly higher registration fee (see registration fees information
on next page).
Further Communications: There will be no further
communications until we receive your pre-registration form. If you need a
personalized invitation, please let us know. We suggest that you make
travel and hotel reservations as soon as possible. Please remember the
October 1, 1994 deadline for pre-registration, and for the title and abstract
of your talk.
Information: If you need further information, please write
or call Jan Want or one of the organizers. We are looking forward to seeing you
in January
of 1995.
Conference Secretary:
Jan Want (409) 845-1269 cat@math.tamu.edu
Center for Approximation Theory
Department of Mathematics
Texas A&M University
College Station, TX 77843-3368
Electronic copies of this conference information, pre-registration form,
and abstract form can be obtained via ftp to wavelet1.math.tamu.edu (USER:
texas8; PASSWORD: texas8), under the file names: texas8.info and texas8.
forms.
From: Charles Chui, Texas A&M University.
Subject: Contents: ACHA, Vol. 1, No. 2
APPLIED AND COMPUTATIONAL HARMONIC ANALYSIS
Time-Frequency and Time-Scale Analysis, Wavelets,
Numerical Algorithms, and Applications (ACHA)
The second issue to be appeared in April 1994 has the following
table of contents.
K. Flornes, A. Grossmann, M. Holschneider, and B. Torresani
"Wavelets on Discrete Fields"
Fritz Keinert
"Biorthogonal Wavelets for Fast Matrix Computations"
N. Hess-Nielsen
"Control of Frequency Spreading of Wavelet Packets"
X. Fang and E. Sere
"Adapted Multiple Folding Local Trigonometric Transforms and Wavelet Packets"
Lars F. Villemoes
"Continuity of Nonseparable Quincunz Wavelets"
Ming-Jun Lai
"On Stromberg's Spline-Wavelets"
Debao Chen
"Extended Families of Cardinal Spline Wavelets"
Jayakumar Ramanathan and Pankaj Topiwala
"Time-Frequency Localization and the Spectrogram"
SUBSCRIPTION RATES:
Institutional Personal
In the U.S.A. and Canada: $184 $74
All other countries: $221 $93
ADDRESS: Academic Press, Inc. 525 B Street, San Diego, CA 92101, U.S.A.
From: publish@baltzer.nl (Daniel Baltzer)
Subject: Contents: Adv. Comput. Math., Volume 2, No. 3, 1994
Advances in Computational Mathematics, Volume 2, No. 3, 1994, ISSN 1019
7168 Editors-in-Chief: John C. Mason (E-mail: j.c.mason@hud.ac.uk) &
Charles A. Micchelli (E-mail: cam@yktvmz.bitnet)
pp 261-277: A parallel implementation of the restarted GMRES iterative
algorithm for nonsymmetric systems of linear equations
R. Dias da Cunha and T. Hopkins
pp 279-317: Time-marching numerical schemes for the electric field integral
equation on a straight thin wire
P.J. Davies and D.B. Duncan
pp 319-341: Acceleration property for the E-algorithm and an application to
the summation of series
M. Prevost
pp 343-355: Hankel operators and best Hankel approximation on the half-plane
X. Li
Requests for FREE SPECIMEN copies and orders for Advances in Computational
Mathematics are to be sent to: E-mail: publish@baltzer.nl
J.C. Baltzer AG, Science Publishers
Asterweg 1A 1031 HL Amsterdam The Netherlands
tel. +31-20-637 0061 fax. +31-20-632 3651
e-mail: publish@baltzer.nl
From: shann@wangwei.math.ncu.edu.tw (Shann Wei-Chang)
Subject: Question: Page 258 of Recent Adv in Wavelet Anal
Page 258 in the two copies of "Recent Advances in Wavelet Analysis", edited
by Schumaker and Webb, that I have examined is missing. It is supposed to
be the references of Daubechies article. Can someone provide that page
electronically on this net?
I know, I should request AP to resolve this problem. But it is inefficient
to do so from Taiwan.
Thank you in advance.
-Shann
Note from the editor:
In an expeditious reply, Larry Schumaker graciously agreed to make the
postscript file of this particular page available. It can be found on our
ftp-site (ftp.math.sc.edu) as the file
/pub/wavelet/papers/varia/page258.ps.
From: ronan@gso.SAIC.COM (Ronan Le Bras)
Subject: Question: Looking for references on wavelets on a sphere.
Can you suggest some references on two-dimensional wavelets defined on
the surface of a sphere (separable or nonseparable).
It seems that such basis, if they exist would provide a good alternative
to spherical harmonics, because of spacial localization.
Thanks . Ronan. ronan@gso.saic.com
From: Yves Misiti <misiti@diamant.matups.fr>
Subject: Question: Exact reconstruction with Daubechies filters ?
In the document Ten Lectures on Wavelets by Ingrid Daubechies, 1992
Tables 8.2 (page 277) and 8.3 (page 279), the coefficients taken for
Synthesis (inverse) filter to be the m_0^tilde filter, seem NOT to
produce exact reconstruction? for DBS(3,7).
Has anybody an answer to this problem? thank you in advance
Yves Misiti
CNRS URA D0743 bat 425 mathematique Universite Paris Sud 91405 Orsay CEDEX
tel : (33) 1 69 41 60 28 fax : (33) 1 69 41 72 34
email : misiti@stats.matups.fr
[Answer] [Answer] [Answer]
From: toet@sunwise.tm.tno.nl (Alexander Toet)
Subject: Book: Shape in Picture
BOOK ANNOUNCEMENT:
Shape in Picture - Mathematical Description of Shape in Grey-level Images
Eds: Ying-Lie O, Alexander Toet, David H. Foster, Henk J.A.M. Heijmans,
Peter Meer
Publisher: NATO ASI Series F, Vol. 126, Springer Verlag, Berlin.
ISBN 3-540-57578-2 Berlin - ISBN 0-387-57578-2 New York
This book contains a selection of papers on different mathematical
approaches to the construction of shape descriptions from grey-level
images. Tutorial sections are included to assist the reader in becoming
acquainted with the variety of approaches to the problem.
TABLE OF CONTENTS
PART 1 MATHEMATICAL BACKGROUND
TOPOLOGY AND GEOMETRY
The Khalimsky Line as a Foundation for Digital Topology, p.3
Ralph Kopperman
Topological Foundations of Shape Analysis, p.21
Vladimir A. Kovalevsky
A New Concept for Digital Geometry, p.37
Vladimir A. Kovalevsky
Theoretical Approaches to N-Dimensional Digital Objects, p.53
Klaus Voss
On Boundaries and Boundary Crack-Codes of Multidimensional Digital Images, p.71
T. Yung Kong
Studying Shape Through Size Functions, p.81
Claudio Uras and Alessandro Verri
CATEGORICAL SHAPE THEORY
Introduction to Categorical Shape Theory,
with Applications in Mathematical Morphology, p.91
Mirek Husek
Shape Theory: an ANR-Sequence Approach, p.111
Jack Segal
Can Categorical Shape Theory Handle Grey-level Images? p.127
Timothy Porter
PART 2 LOCAL EXTRACTION
MATHEMATICAL MORPHOLOGY
Mathematical Morphology as a Tool for Shape Description, p.147
Henk J.A.M. Heijmans
On Information Contained in the Erosion Curve, p.177
Juliette Mattioli and Michel Schmitt
Morphological Area Openings and Closings for Grey-scale Images, p.197
Luc Vincent
Manifold Shape: from Differential Geometry to Mathematical Morphology, p.209
Jos B.T.M. Roerdink
On Negative Shape, p.225
Pijush K. Ghosh
WAVELETS
An Overview of the Theory and Applications of Wavelets, p.249
Bjorn Jawerth and Wim Sweldens
Fractal Surfaces, Multiresolution Analyses, and Wavelet Transforms, p.275
Jeffrey S. Geronimo, Douglas P. Hardin, and Peter R. Massopust
Interpolation in Multiscale Representations, p.291
Charles H. Anderson and Subrata Rakshit
PART 3 THEORY OF SHAPE
KEYNOTE
Discrete Stochastic Growth Models for Two-Dimensional Shapes, p.301
Scott Thompson and Azriel Rosenfeld
DIFFERENTIAL GEOMETRY
Classical and Fuzzy Differential Methods in Shape Analysis, p.319
David H. Foster
Elements of a Fuzzy Geometry for Visual Space, p.333
Mario Ferraro and David H. Foster
On the Relationship Between Surface Covariance and Differential Geometry, p.343
Jens Berkmann and Terry Caelli
Image Representation Using Affine Covariant Coordinates, p.353
Jun Zhang
THEORY OF SHAPE PERCEPTION
Equivariant Dynamical Systems:
a Formal Model for the Generation of Arbitrary Shapes, p.363
William C. Hoffman
Neural Processing of Overlapping Shapes, p.383
Andre J. Noest
Contour Texture and Frame Curves for the Recognition of Non-Rigid Objects, p.393
J. Brian Subirana-Vilanova
PART 4 SYMBOLIC REPRESENTATION
SHAPE PRIMITIVES
Conic Primitives for Projectively Invariant
Representation of Planar Curves, p.403
Stefan Carlsson
Blind Approximation of Planar Convex Shapes, p.415
Michael Lindenbaum and Alfred M. Bruckstein
Recognition of Affine Planar Curves Using Geometric Properties, p.423
Craig Gotsman and Michael Werman
Recognizing 3-D Curves from a Stereo Pair of Images:
a Semi-differential Approach, p.433
Theo Moons, Eric J. Pauwels, Luc J. Van Gool, Michael H. Brill,
and Eamon B. Barrett
Statistical Shape Methodology in Image Analysis, p.443
John T. Kent and Kanti V. Mardia
Recognition of Shapes from a Finite Series of Plane Figures, p.453
Nikolai M. Sirakov
Polygonal Harmonic Shape Characterization, p.463
Anthony J. Maeder, Andrew J. Davison, and Nigel N. Clark
Shape Description and Classification Using the Interrelationship of
Structures at Multiple Scales, p.473
Gregory Dudek
Learning Shape Classes, p.483
Stanley M. Dunn and Kyugon Cho
HIERARCHICAL REPRESENTATION
Inference of Stochastic Graph Models for 2-D and 3-D Shapes, p.493
Jakub Segen
Hierarchical Shape Analysis in Grey-level Images, p.511
Annick Montanvert, Peter Meer, and Pascal Bertolino
Irregular Curve Pyramids, p.525
Walter G. Kropatsch and Dieter Willersinn
Multiresolution Shape Description by Corners, p.539
Cornelia Fermueller and Walter G. Kropatsch
Model-based Bottom-Up Grouping of Geometric Image Primitives, p.549
Peter Nacken and Alexander Toet
Hierarchical Shape Representation for Image Analysis, p.559
O Ying-Lie
PART 5 EVOLUTIONARY SYSTEMS
EVOLUTIONARY REPRESENTATION
Scale-Space for N-dimensional Discrete Signals, p.571
Tony Lindeberg
Scale-Space Behaviour and Invariance Properties of Differential
Singularities, p.591
Tony Lindeberg
Exploring the Shape Manifold: the Role of Conservation Laws, p.601
Benjamin B. Kimia, Allen R. Tannenbaum, and Steven W. Zucker
Performance in Noise of a Diffusion-based Shape Descriptor, p.621
Murray H. Loew and Sheng-Yuan Hwang
Towards a Morphological Scale-Space Theory, p.631
Rein van den Boomgaard and Arnold W.M. Smeulders
Geometry-based Image Segmentation Using Anisotropic Diffusion, p.641
Ross T. Whitaker and Stephen M. Pizer
MULTISCALE DESCRIPTION
Images: Regular Tempered Distributions, p.651
Luc M. J. Florack, Bart M. ter Haar Romeny, Jan J. Koenderink,
and Max A. Viergever
Local and Multilocal Scale-Space Description, p.661
Alfons H. Salden, Bart M. ter Haar Romeny, and Max A. Viergever
From: shann@wangwei.math.ncu.edu.tw (Shann Wei-Chang)
Subject: Preprint: Quadratures Involving Polynomials and Daubechies' Wavelets
I probabely have announced this before. But I have several paragraphs
revised. It is Technical Report 9301 of Math Dept, National Central
Univ, Taiwan. Anonymous ftp available at
dongpo.math.ncu.edu.tw:/pub/shann/publications/9301.ps
Coauthor: Mr. Jann-Chang Yan
Abstract:
Scaling equations are used to derive formulae of quadratures involving
polynomials and scaling/wavelet functions with compact supports; in
particular, those discovered by Daubechies. It turns out that with a
few parameters, which are theoretically exact, these quadratures can be
evaluated with algebraic formulae instead of numerical approximations.
Those parameters can be obtained with high precision by solving
well-conditioned linear systems of equations which involve matrices
already seen in the literature of wavelets for other purposes.
Wei-Chang Shann, Associate Professor
Dept of Mathematics, National Central University, Chung-Li, Taiwan, R.O.C.
(03)425-6704 ext. 166 (work); (03)425-7379 (work.FAX)
shann@math.ncu.edu.tw
From: Marilyn Grossman, Texas A&M University.
Subject: Course: Wavelets for Engineering Applications
Wavelets for Engineering Applications,
August 17-20, 1994, Texas A&M University in College Station, Texas.
Applications-oriented short course featuring Charles K. Chui, author
of the popular text, "An Introduction to Wavelets," and two experts of
the wavelets research team at Texas A&M University, A.K. Chan
(electrical engineering) and P.K. Yuen (computer science). Designed to
build on your basic understanding of wavelet theory, this course will
take you through the concept of time-frequency analysis, the
significance of the integral wavelet transform, and real-time
preprocessing of a continuous signal using cardinal splines. You will
learn how to construct wavelets, how to use decomposition and
reconstruction algorithms, and how to use wavelet packets for frequency
domain fine-tuning in our hands-on laboratory sessions.
For registration information or to request a brochure, please
contact Marilynn Grossman: email: drgrossm@teex.tamu.edu
Ph: 800/447-9470 or 409/862-4615
Fax: 409/845-5726
From: rads@hartley.gsfc.nasa.gov (Raadhakrishnan Poovendran)
Subject: Question: Looking for a particular function
I am looking for a function f(x) that will satisfy the following
conditions:
1. f(2^m x -n) when multiplied by f(2^m1 x -n1) and integrated
with respect to x leads to
1 if m=m1, n=n1
0 else
2. f(2^m x -n), f(2^m1 x -n1) and f(2^m2 x -n2) when multiplied
and integrated lead to
1 if m=m1=m2 && n=n1=n2
0 else.
Can somebody help me?
From: Tilo Schuerer <tilo@cs.tu-berlin.de>
Subject: Question: Wavelets in Speech Recognition
First test: WAVELETS in Speech Recognition
In the last time I read much about usage of wavelets in signal compression.
My idea is that if I can compress a signal then the computed coefficients after
the compression can be used for classification too because they must
represent essential information about the original signal.
I made some tests in the field of speech recognition in order to
verify that. This are my results after 2 weeks of work:
o According to the paper from Neil Getz
"A Fast Discrete Periodic Wavelet Transform"
Memorandum No. UCB/ERL M92/138
I made some test in compression of speech signals. I tried to find out
which levels of the computed detail-coefficients of the FPWT
are necessary when compressing and decompressing some speech data.
When using a window with 256 points and no overlap the 2nd, 3rd and 4th
level detail-coefficients (56 values) are enough to reconstruct the
speech data so that you can still well understand what was said.
o My first test was in the field of isolated word recognition over telephone
lines. I recorded around 200 calls of people speaking german digits
and some controlling words (approx. 3500 words total). With conventional
methods (PLP, RastaPLP, MFCC and LPC) together with VQ and an MLP I can
get a recognition performance of around 89% over all data.
Now I substituted only the feature extraction and used the detail-
coefficients of the FPWT from level 2 to level 4 (56 values total).
But I only reached 35% recognition performance of the unknown test data
while the training data was recogniced to 80%. I think it is clear that
I will not be better than the conventional methods but I never thought
I would be so bad!
Therefore my questions:
o Did anybody make some tests in using wavelets in speech recognition?
o Does it really make sense to use the detail-coefficients of the FPWT?
o I read a lot about scalograms in order to show the energy of wavelet-
coefficients in the time-sequency-plane. This would be very similar
to spectrograms derived from the FFT. How can I compute the scalogram?
Thanks in advance for your help and best greetings,
Tilo Schuerer
From: Marilyn Radcliff <radcliff@math.ohio-state.edu>
Subject: Contents: JAT Vol. 76 No.3, March. 94
Journal of Approximation Theory, Volume 76, Number 3, March 1994
Table of Contents
Maoli Chang. The Behavior of Polyharmonic Cardinal Splines as Their Degree
Tends to Infinity, 287-302.
Charles K. Chui and Jian-Zhong Wang. Quasi-Interpolation Functionals on
Spline Spaces, 303-325.
M. Finzel. Linear approximation in $\ell_n^\infty$, 326-350.
S. Belmehdi and A. Ronveaux. Laguerre-Freud's Equations for the Recurrence
Coefficients of Semi-classical Orthogonal Polynomials, 351-368.
Vilmos Totik. Strong Converse Inequalities, 369-375.
Jarom\'ir \u Sim\u sa. Approximations of Two-Place Functions by Finite Sums
of Terms with Separable Variables, 376-392.
Ding-Xuan Zhou. Weighted approximation by Sz\'asz--Mirakjan operators,
393-402.
Ding-Xuan Zhao. Weighted Approximation by Multidimensional Bernstein
Operators, 403-422.
Note
Kazuaki Kitahara. Weak Tchebycheff Systems, 423-425.
Marilyn Radcliff radcliff@math.ohio-state.edu
Department of Mathematics
The Ohio State University
From: Akbar Safavi <akbar@ce.su.oz.au>
Subject: Question: Non-Separable Multi-Dimensional Wavelets
Hi Everybody
I am wondering if there is any software for non-separable
multi-dimensional wavelets available to the public. I need
such a software for wavelets applications. Could anybody
help me on this matter?
regards
akbar
From: tsuhan@big.att.com (Tsuhan Chen)
Subject: Answer: Multidimensional Factorization (WD 3.3 #9)
Dear David,
Some interesting work on parameterization of multidimensional (MD)
filter banks and factorization of lossless matrices of multivariables
can be found in:
Sankar Basu, Han-Mook Choi, and C. H. Chiang, "On non-separable
multidimensional perfect reconstruction filter banks," Conf. Record of
27th Annual Asilomar Conference on Signals, Systems, and Computers,
vol. 1, pp. 45-49, Pacific Grove, CA, November 1993.
and the references therein.
Sincerely,
Tsuhan Chen
AT&T Bell Laboratories, 4C528 101 Crawfords Corner Road Holmdel, NJ 07733-3030
Email: tsuhan@research.att.com
From: Riverside Research Institute <rribro@world.std.com>
Subject: Question: Looking for real-time application of wavelets
I am trying to locate paper(s) on a "real-time" application of the
wavelet transform, i.e., implementation of a real-time system using the
wavelet transform. Also, trying to find out what the computational
burden of implementing the transform real-time, n*log(n) is well known
for the fourier transform.
Appreciate any comments, papers, etc...
robert l hernandez
rribro@world.std.com
<webmaster@wavelet.org>