[Home] [Back Issues] [Search] [Calendar] [Vol 1] [Vol 2] [Vol 3] [Vol 4] [Vol 5] [Vol 6] [Vol 7] [Vol 8] [Vol 9]

The Wavelet Digest, Volume 9, Issue 3 (April 28, 00)




Today's Editor: Wim Sweldens
                Lucent Technologies, Bell Laboratories
                http://wim.sweldens.com



Today's Topics:

     1: Preprint: Fourier transform of Hermite interpolatory subdivision schemes
     2: Preprint: Wavelets for adaptive solution of boundary value problems
     3: Preprint: A compound model of TCP connection arrivals
     4: Preprint: Efficient encoding of the significance maps..
     5: Preprint: A Wavelet Method for Film-Grain Noise Corrupted Images
     6: Preprint: Fast Iterative Solution of Saddle Point Problems
     7: Thesis:   Combined Subdivision Schemes (A. Levin)
     8: Thesis:   Wavelet Analysis for Detection of Rail Irregularities
     9: Thesis:   wavelet thresholding and noise reduction (M. Jansen)
    10: Software: A wavelet based image coder
    11: Web:      A web page on multiscale skeletons
    12: Meeting:  The Eleventh British Machine Vision Conference
    13: Meeting:  Seminar "Wavelets and their Applications".
    14: Course:   Video Compression Short Course in Portland 8/7-12/2000
    15: Course:   Wavelets, Image Analysis, and Lossless Coding
    16: Job: PhD in Switzerland: Wavelets and splines in medical imaging
    17: Job:      Research Studentships at Sheffield University, UK
    18: Cfp:      Special issue on subdivision surfaces (CAGD)
    19: Contents: J. Approx. Th. - apr00 TOC
    20: Question: Nasa comparison of wavelet based video codecs
    21: Question: Performance of Wavelet based applications
    22: Question: Spherical wavelets
    23: Question: Wavelets and Stereo vision
    24: Question: Wavelets and impulse-type signals
    25: Question: Wavelets for classification
    26: Question: Estimation of Lipschitz exponent 
    27: Question: Music Recognition & wavelets
     
Current number of members: 15801

E-mail:
   General help:      help@wavelet.org
   Add yourself:      add@wavelet.org
   Remove yourself:   remove@wavelet.org with Subject: $ADDRESS
   Publish in the WD: publish@wavelet.org 
   Publishing policy: policy@wavelet.org

Visit us on the Web at http://www.wavelet.org/

Calendar of events:

 Mar 6-10:  Numerical Methods in Multiphase Flow, Germany
 May 2-6: Wavelets 2000, Havana, Cuba
 Mar 12-16: Australian Society for Exploration Geophysics, Perth
*May 17:    "Wavelets and their Applications", CWI, Amsterdam
 May 17-20: Trends in Approximation Theory 2000, Nashville, TN
 May 22-26: CBMS-NSF Regional Conference, St-Louis
 May 29-31: Gabor centenary conference, Vienna
 Jun 5-9:  Wavelets at the SIMAI 2000 conference, Ischia, Italy
 Jun 29 - Jul 4: 5th Curves and Surfaces, Oslo, Norway
 Jul 13-17: Steve Smale's 70th birthday, Hong Kong
 Jul 17-19: Wavelet Applications, Tokyo, Japan
 Jul 30 - Aug 2: SPIE Wavelet Applications VIII, San Diego
 Jul 30 - Aug 2: Lifting session at the SPIE meeting,  San Diego
 Jul 30 - Aug 2: IEEE Multimedia And Expo, New York
*Aug 7-12:  Video Compression Short Course, Portland, OR
*Sep 11-14: The Eleventh British Machine Vision Conference Bristol
 Sep 22-24: AMS Special Session on Wavelets, Toronto
*Nov 13-16: Wavelets, Image Analysis, and Lossless Coding







Date: Tue, 28 Mar 2000 18:15:40 -0500
From: Daniel Lemire <Daniel.Lemire@videotron.ca>
Subject: #1 Preprint: Fourier transform of Hermite interpolatory subdivision schemes

Title:
Fourier transform of Hermite interpolatory subdivision schemes

Authors and their email:
Serge Dubuc: dubucs@dms.umontreal.ca
Daniel Lemire: lemire@ondelette.com
Jean-Louis Merrien: Jean-Louis.Merrien@insa-rennes.fr

Abstract
Two subdivision schemes with Hermite data on Z are studied.
These schemes use 2 or 7 parameters respectively depending on
whether Hermite data involve only first derivatives or include
second derivatives. For a large region in the parameters space,
the schemes are $C^1$ or $C^2$ convergent or at least are
convergent on the space of Schwartz distributions. The Fourier
transform of any interpolating function can be computed through
products of matrices of order 2 or 3. The Fourier transform is
related to a specific system of functional equations whose
analytic solution is unique except for a multiplicative constant.
The main arguments for these results come from
Paley-Wiener-Schwartz theorem on the characterization of the
Fourier transforms of distributions with compact support and a
theorem of Artzrouni about convergent products of matrices.

Keywords : Hermite interpolation, curve fitting,
subdivision, Fourier transform, distributions, convergence of
infinite products, products of matrices.

To download a PostScript file :

http://www.ondelette.com/ETRANSF.html



Date: Wed, 29 Mar 2000 15:37:52 +0200
From: Jean-Baptiste Caillau <Jean-Baptiste.Caillau@enseeiht.fr>
Subject: #2 Preprint: Wavelets for adaptive solution of boundary value problems

"Wavelets for adaptive solution of boundary value problems"
J. B. Caillau and J. Noailles
Accepted, 16th IMACS World Congress 2000, Lausanne,  Switzerland,
August 2000.

Keywords.  boundary  value  problems,  adaptive   discretization,
wavelet   analysis,   multiple  shooting  and  finite  difference
methods, nonlinear optimal control

AMS subject classifications. 65L10, 65L50, 42C99, 49M05

Abstract. Among the classical methods for the numerical  solution
of  boundary value problems with ordinary differential equations,
we can distinguish several parametrization levels of the underly-
ing  time  discretizations: while spectral (Vlassenbroeck (1988))
or  pseudo-spectral  (El-Gindy  (1994))   techniques   are   only
parametrized  by  the  number of points used for the quadratures,
finite difference or more generally multiple shooting  approaches
depend  on up to three parameters: the number of shooting points,
the number of steps on each shooting interval, and the number  of
points  used  for the quadrature at each step. Modern integrators
control the step as well as the quadrature degree, but usually do
not provide a mechanism to choose the shooting points themselves.
In this regard, we propose an adaptive  discretization  technique
based  on a wavelet analysis (see also Jameson (1996)) that takes
advantage of the information currently available on  the  problem
being  solved to define dynamically a finite difference or shoot-
ing mesh.  Depending on the problem, several  couplings  of  this
adaptive  discretization  with  the method itself (finite differ-
ences, multiple shooting...) are  proposed:   coupling  with  the
nonlinear  iterations,  coupling with solution (the BVP is solved
repeatedly on more and more adapted meshes to gain accuracy),  or
even coupling with a continuation procedure. We make use of these
ideas on linear and nonlinear  BVPs.  In  particular,  a  problem
arising from optimal control in celestial mechanics is studied.

URL of the file:
http://www.enseeiht.fr/apo/caillau/papers/imacs2000.html

J. B. Caillau
ENSEEIHT-IRIT UMR CNRS 5505 2, rue Camichel 31071 Toulouse France
caillau@enseeiht.fr
http://www.enseeiht.fr/apo/caillau



Date: Tue, 11 Apr 2000 08:29:24 +0300
From: "Wim Sweldens" <wim@bell-labs.com>
Subject: #3 Preprint: A compound model of TCP connection arrivals

Title: A Compound Model for TCP Connection Arrivals

Authors: Carl J. Nuzman, Iraj Saniee, Wim Sweldens, Alan Weiss 

Abstract: We propose a two level model for TCP connection arrivals in
local area networks. The first level are user sessions whose arrival
is time-varying Poisson. The second level are connections within a
user session. Their number and mean interarrival are independent and
biPareto across user session. The interarrivals within a user session
are Weibull, and across all users are correlated Weibull. Our model
has a small number of parameters which are inferred from real traffic
collected a a firewall. We show that traffic synthesized with our
model closely characterizes the original data.

Status: Technical Memorandum, Bell Laboratories, Lucent Technologies,
March 2000. 

Available on the web at /who/wim/papers/tcpmodel/



Date: Mon, 10 Apr 2000 09:27:24 +0300
From: "Rusen Oktem" <rusen@cs.tut.fi>
Subject: #4 Preprint: Efficient encoding of the significance maps..

Levent Oktem, Rusen Oktem, Karen Eguiazarian, Jaakko Astola

"Efficient encoding of the significance maps in wavelet based image
compression", to appear in Proc. of ISCAS 2000, 28-31 May, Geneva,
Switzerland

Abstract: We propose a method for the efficient encoding of the
significance maps in wavelet-based image compression. Adopting the
significance map encoding part of the morphology-based MRWD, we
introduce three new features: exploitation of the knowledge of
coefficient magnitudes in the parent band; reordering of the residual
part of the significance map; and the employment of hierarchical
enumerative coding, a new entropy coding method, instead of arithmetic
coding. Experimental results show that utilization of these features
bring a consistent improvement. The average improvement is around
7.6%.

The pdf file of the full paper can be found at
http://www.cs.tut.fi/~rusen/sigmap.pdf

Rusen Oktem
Signal Processing Laboratory,
Tampere University of Technology,
P.O. Box 553, 33101 Tampere, Finland
Tel: +358 400 830641 Fax: +358 3 365 3857
e-mail: rusen@cs.tut.fi
web: http://www.cs.tut.fi/~rusen



Date: Mon, 10 Apr 2000 09:31:51 +0300
From: "Rusen Oktem" <rusen@cs.tut.fi>
Subject: #5 Preprint: A Wavelet Method for Film-Grain Noise Corrupted Images

Rusen Oktem, Karen Egiazarian
"A Wavelet Transform Method for  Coding Film-Grain Noise Corrupted Images",
to appear in Proc. of ICASSP 2000, 5-9 June, Istanbul, Turkey.

Abstract: We propose an algorithm for compressing images corrupted by
film-grain type noise.  The algorithm nonlinearly modifies the
transform coefficients of noisy data for reducing the impact of noise
on compression.  Although any orthogonal transform is suitable for the
application, orthogonal wavelet transform is preferred for simplicity
in calculating the filter coefficients and for obtaining better
compression performance. The algorithm is tested on different images
at different noise variances and the results are compared to spatial
filtering before compression. Our method outperforms spatial
prefiltering in terms of PSNR values and it is easier to implement.

The pdf file of the full paper can be found at
http://www.cs.tut.fi/~rusen/icassp2000.pdf

Rusen Oktem
Signal Processing Laboratory,
Tampere University of Technology,
P.O. Box 553, 33101 Tampere, Finland
Tel: +358 400 830641 Fax: +358 3 365 3857
e-mail: rusen@cs.tut.fi
web: http://www.cs.tut.fi/~rusen



Date: Mon, 10 Apr 2000 11:20:01 +0200
From: Angela Kunoth <kunoth@iam.uni-bonn.de>
Subject: #6 Preprint: Fast Iterative Solution of Saddle Point Problems

    Fast Iterative Solution of Saddle Point Problems 
           in Optimal Control Based on Wavelets

                       Angela Kunoth

Abstract: For the numerical solution of an elliptic boundary value
problem with boundary control, the problem is formulated as minimizing
a quadratic functional involving natural norms of the state and the
control.  Firstly the constraint, the elliptic boundary value problem,
is formulated in an appropriate weak form that allows to handle
varying boundary conditions explicitly. Namely, the boundary
conditions are appended by Lagrange multipliers, leading to a saddle
point problem. This is combined with a fictitious domain approach in
order to cover also more complicated boundaries.

>From the optimality conditions for the minimization problem, a second
saddle point problem is derived. It is shown by standard techniques
that the resulting weakly coupled system of the two saddle point
problems admits a unique solution. For its discretization,
(biorthogonal) wavelets are used which allows to formulate the matrix
of the coupled system as an $\ell_2$--operator, initially in an
infinite--dimensional space.  Using an appropriate least--squares
formulation, the stable invertibility of the discrete
finite--dimensional coupled system matrix is shown. This matrix also
has a uniformly bounded condition number independent of the
discretization.

In order to solve the resulting linear system of coupled saddle point
problems, different iterative methods are proposed. One is to use a
conjugate gradient method applied to the normal equations stemming
from the least squares approach. As an alternative, a fully iterative
method is described which consists of an outer iteration which
alternatingly solves the two saddle point problems, and an inner
iteration to solve each of the saddle point problems, exemplified in
terms of the Uzawa algorithm. It is proved here that this strategy
converges, provided that the inner systems are solved sufficiently
well.  Moreover, since the system matrix is well--conditioned, it is
shown that in combination with a nested iteration strategy this
iteration is asymptotically optimal in the sense that it provides the
solution on discretization level $J$ with an overall amount that is
proportional to the number of unknowns $N_J$ on that level.

Finally, numerical results are provided.         

http://www.iam.uni-bonn.de/~kunoth/papers/is.ps.gz

Angela Kunoth
Institut fuer Angewandte Mathematik
Universitaet Bonn
Wegelerstr.6
53115 Bonn, Germany



Date: Fri, 31 Mar 2000 17:55:16 +0200
From: Adi Levin <adilev@inter.net.il>
Subject: #7 Thesis: Combined Subdivision Schemes (A. Levin)

A postscript version of the thesis is available through

http://www.math.tau.ac.il/~adilev

In this dissertation, we introduce a new class of subdivision schemes
called Combined Subdivision Schemes in an attempt to solve some of the
difficulties that subdivision schemes present to designers of
mechanical parts.  In the mechanical design industry, the continuity
and smoothness of the mathematical model is crucial to its visual
quality and to the successful application of accurate analysis and
simulation algorithms.  An object is typically represented by a
collection of surfaces that surround it.  Thus, the ability of the
mathematical model of these surfaces to accurately satisfy boundary
conditions has a considerable effect on the quality of the
mathematical representation of the geometric object.

Combined subdivision schemes allow to prescribe arbitrary boundary
conditions, such that the generated smooth surfaces will satisfy them
exactly.  The boundary conditions can include boundary curves as well
as partial derivatives of arbitrary order of the surface at the
boundary. The same principles that we use for satisfying boundary
conditions can be applied to the interpolation of a net of curves by a
surface.  Every combined subdivision scheme operates like an ordinary
subdivision scheme with special subdivision rules near the boundary or
near the interpolated curves. The theory deals with necessary and
sufficient conditions for the generated surfaces to have desired
smoothness and approximation properties.

This dissertation is composed of two parts. Part I develops the theory
of combined subdivision schemes - their definition, and the analysis
of their smoothness and approximation properties.  In part II methods
for constructing new schemes based on the theory are
presented. Several combined schemes are derived and their performance
is demonstrated by examples.



Date: Tue, 11 Apr 2000 11:30:12 +0200
From: "Machado, G., R." <T96_MAM@t.kth.se>
Subject: #8 Thesis: Wavelet Analysis for Detection of Rail Irregularities

Master Thesis:

Title:	"Wavelet Analysis and its Application for Detection of Track 
            and Rail Irregularities".

Author:	Machado, G., R.

Pages:	84 

Institution:	Department of Vehicle Engineering

			The Marcus Wallenberg Laboratory for Sound and 		
			Vibration Research (MWL),

			Royal Institute of Technology, Sweden

Contents:

		1- Introduction (1)                                         

		2- Wavelet Theory. A brief discussion (8)

		3- Filters and Filter Banks (32)

		4- Wavelts and Wavelt Packets in Practice(46)

		5- Detecting Track and Rail Irregularities(63)

		6- Conclusions(70)

		Appendix A (71)

		Bibliography (74)

Notes:	No previous knowledge about the subject is required.<FontFamily><param>Times New Roman</param> <FontFamily><param>Arial</param>

Copies might be obtained by asking some of the following staff: 

Anders Nilsson
andersni@fkt.kth.se

Hans Boden
hansb@fkt.kth.se

Yours sincerely

Machado, G., R.



Date: Tue, 11 Apr 2000 12:37:47 +0200 (MET DST)
From: Maarten Jansen <Maarten.Jansen@cs.kuleuven.ac.be>
Subject: #9 Thesis: wavelet thresholding and noise reduction (M. Jansen)

Title: Wavelet thresholding and noise reduction

Author: Maarten Jansen

Supervisor: Prof. Adhemar Bultheel

Institution: K.U.Leuven, department of computer science

Date: April, 7, 2000

Keywords: 

wavelet, noise, threshold, non-linear, non-parametric regression, generalized
cross validation, minimum risk, mean square error, Bayes, Markov Random Field,
non-equidistant

Abstract:

A wavelet transform decomposes data into a sparse, multiscale
representation.  This dissertation uses both features in wavelet based
noise reduction algorithms.

Sparsity is the key to wavelet thresholding: coefficients with
magnitude below a threshold are replaced by zero. After an
introduction to wavelets and their applications, this text discusses
the minimum risk threshold. This threshold minimizes the expected mean
square error of the output. This error cannot be computed exactly when
the uncorrupted data are unknown. We present a procedure based on
generalized cross validation (GCV) to estimate the optimal
threshold. An asymptotic argument motivates this estimation method. To
this end, we first study the asymptotic behavior of the minimum risk
threshold.  We compare these minimum risk and GCV thresholds with the
well known universal threshold.

The multiresolution character of a wavelet decomposition allows for
refinements of the general threshold scheme. Tree structured
thresholding reduces false structures in the output. Scale dependent
thresholds are necessary to deal with correlated noise or
non-orthogonal wavelet transforms. The synthesis from a non-decimated
wavelet transform has an additional smoothing effect. We also
investigate noise reduction in the framework of integer wavelet
transform.

The next part concentrates on images. An approximation theoretic
argument learns that wavelets might not be the ultimate basis
functions for image processing. Moreover, selecting the coefficients
with large magnitude is a local approach. A Bayesian procedure could
lead to a more structured coefficient selection, which better
preserves edges in the output. The geometric prior model favors
clusters of important coefficients.

The last part investigates the applicability of threshold algorithms
for non-equidistant data and second generation wavelets. Experimental
results indicate that instability of the wavelet transform hinders the
classification of the coefficients according to their importance. We
propose an algorithm to overcome this difficulty.

website: (with thesis in compressed postscript)

http://www.cs.kuleuven.ac.be/~maarten/publications/PhD/index.html 

Maarten Jansen
http://www.cs.kuleuven.ac.be/~maarten



Date: Mon, 17 Apr 2000 12:40:11 +0300 (EET DST)
From: "Oktem Levent" <loktem@cs.tut.fi>
Subject: #10 Software: A wavelet based image coder

Our wavelet-based image coder can be downloaded from 
http://www.cs.tut.fi/~loktem/umut/index.html

Description of the algorithm and its performance table are also
included in the same page.

Levent Oktem
Signal Processing Laboratory,
Tampere University of Technology,
P.O. Box 5512, 33101 Tampere, FINLAND

Tel: +358 40 707 7346
e-mail: loktem@cs.tut.fi
 


Date: Thu, 6 Apr 2000 22:42:28 -0300 (EST)
From: Luciano Da Fontoura Costa <luciano@if.sc.usp.br>
Subject: #11 Web: A web page on multiscale skeletons

7th April 2000

We are glad to inform that a homepage has now been set describing what
is possibly the simplest approach to obtaining high quality multiscale
skeletons.  The method, based on the concept of exact dilations,
allows the simultaneous determination of the exact Euclidean distance
transform and exact label propagation.  The obtained family of
multiscale skeletons present the distinctive feature that a series
reconstructions of the shape can be obtained where small scale
information is removed while preserving the position of larger scale
details in the shape.  The applications of this methodology includes
shape analysis, biology, material sciences, line drawing analysis,
visual perception, and many more.

The page, which can be found at

       http://cyvision.if.sc.usp.br/msskeletons/

includes a description of the methods, pseudo-codes, animations,
a downloadable interactive demo, and references.

Comments and suggestions are very welcomed.

Best wishes, Luciano

Prof. Luciano da Fontoura Costa
Coordinator - Cybernetic Vision Research Group
DFI-IFSC, Universidade de Sao Paulo
Caixa Postal 369
Sao Carlos, SP
13560-970
Brazil
FAX: +55 162 73 9879  or  +55 162 71 3616

e-mail:  
   luciano@if.sc.usp.br
Group homepage:
   http://cyvision.if.sc.usp.br/
Personal homepage: 
   http://cyvision.if.sc.usp.br/~luciano



Date: Tue, 21 Mar 2000 09:21:13 +0000
From: Majid Mirmehdi <M.Mirmehdi@compsci.bristol.ac.uk>
Subject: #12 Meeting: The Eleventh British Machine Vision Conference

                  B M V C  2 0 0 0
     The Eleventh British Machine Vision Conference
               University of Bristol
                11-14 September 2000
       http://www.cs.bris.ac.uk/Events/BMVC2000/

       C  A  L  L     F  O  R     P  A  P  E  R  S

The British Machine Vision Conference (BMVC) is the main UK 
conference on machine vision and related areas.  Organised by 
the British Machine  Vision Association, the 11th BMVC will be 
held in September 2000 at the University of Bristol, Bristol, 
UK. Papers will be refereed on their originality, presentation, 
empirical results, and quality of evaluation. Conference topics 
include (but are not limited to):

 * image features and coding     * object recognition
 * multi-view and stereo vision  * real-time and active vision
 * grouping and segmentation     * shape and surface geometry
 * texture, shading and colour   * medical and industrial appls.

BMVC2000 will be a single-track meeting with oral and poster 
presentations. Invited speakers are Prof. Eric Grimson (MIT) and 
Prof. Luc Van Gool (KUL). The Proceedings will be available to 
delegates at the conference in hard copy and on CD, and a selection
of the best papers will be published separately in a special issue 
of the journal Image and Vision Computing. There will also be a 
pre-Conference Tutorial programme presented by Prof. Andrew 
Blake (Microsoft).

As in previous years, BMVC 2000 will incorporate a designated 
Industry Day on Wednesday 13th September. Presentations on Industry
Day will focus on the practical applications of machine vision, and 
we are particularly keen to see presentations of well evaluated 
applications which clearly show thorough understanding of the 
underlying principles. Delegates will be able to view poster 
presentations and see demonstrations by both industrial exhibitors
and researchers demonstrating their academic work. Industry Day 
is organised with the co-operation of the UK Industrial Vision 
Association. 

The Applied Vision Association is also holding its 3rd one day 
Meeting on Natural Images at the University of Bristol on Friday
Sept 15th 2000. This is being arranged as a "satellite" meeting 
to the BMVC2000 conference. For those who wish to submit a paper
for this meeting or to register and attend details are available 
on the BMVC2000 WWW page. 

                         IMPORTANT DATES

Deadline for paper submission:           April 14th 2000
Notification of acceptance:              June  16th 2000
Deadline for camera-ready copy:          July   7th 2000

CONFERENCE CO-CHAIRS:                  CONFERENCE SECRETARIAT:
Dr. Majid Mirmehdi                      Mrs. Ann Prowse
M.Mirmehdi@cs.bris.ac.uk                A.Prowse@cs.bris.ac.uk
Prof. Barry Thomas                      Tel: +44 (0)117 954 5130
B.Thomas@cs.bris.ac.uk                  Fax: +44 (0)117 954 5208
Department of Computer Science, University of Bristol, 
Merchant Venturers Building, Woodland Road, Bristol BS8 1UB, UK

Papers will be accepted either for oral presentation or as posters. 
By submitting a paper to BMVC, the authors warrant that it, and 
any related paper with essentially the same technical content, has
not been and will not be submitted to any other conference during 
the BMVC review period. Full papers, not exceeding 10 printed pages
(approx. 5000 words if there are no figures), may be submitted 
electronically. For further information and paper submission 
guidelines please either see the BMVC2000 WWW pages at:

http://www.cs.bris.ac.uk/Events/BMVC2000/
or contact Majid Mirmehdi using the details above.  

BMVC2000 is part-sponsored by HP Research Laboratories Bristol.

The British Machine Vision Association and Society for Pattern 
Recognition is a Company limited by guarantee No. 2543446, registered
in England and Wales. Registered Office: Granta Lodge, 71 Graham Road,
Malvern, Worcs. WR14 2JS. The BMVA is a non-profit-making body and is 
registered as charity No. 1002307.
  


Date: Tue, 28 Mar 2000 14:55:47 +0200
From: "L. Kamstra" <Lute.Kamstra@cwi.nl>
Subject: #13 Meeting: Seminar "Wavelets and their Applications".

Seminar "Wavelets and their Applications"

In the framework of the NWO research project "Wavelets and their
Applications" in which CWI, RUG, TUE and UT participate, a general
seminar dealing with various aspects of theoretical and applied
aspects of wavelets and related areas, will be organised twice a year
(May/June and November/December). Each meeting will consist of two
parts. The morning session consists of an invited two-hour lecture by
one of the leading experts in wavelet theory. The afternoon session,
comprising two or three lectures will be devoted to various topics.

The first meeting will be held on the 17th of May 2000 in room Z011 at
CWI in Amsterdam. Main speaker is

      Wim Sweldens from Lucent Technologies.

The full programme will be announced on the following website:

      http://www.cwi.nl/projects/wavelets/

There one can also find other information concerning the NWO project
and the seminar and fill out a registration form (note that
registration is free of charge). Further information can also be
obtained from Lute Kamstra, Lute.Kamstra@cwi.nl, tel (+31) 20 592 4239.

Lute.Kamstra@cwi.nl
CWI department PNA4
Office M501
Phone (+31) 20 592 4239



Date: Mon, 20 Mar 2000 17:03:57 -0800 (PST)
From: Fu Li <fli@ee.pdx.edu>
Subject: #14 Course: Video Compression Short Course in Portland 8/7-12/2000

4 & 1/2 Days (August 7 - 12, 2000) in Portland Oregon

                                on

                IMAGE AND VIDEO COMPRESSION:
        FUNDAMENTALS, STANDARDS, AND APPLICATIONS

                        Featured:

 Majid Rabbani -                        Eastman Kodak
 M. Ibrahim Sezan -                     Sharp Labs. of America
 Thomas Gardos -                        Intel Corporation

Organized by Fu Li
Portland State University, Portland, Oregon, USA

                please visit

http://www.ee.pdx.edu/~compress

Contact Don Mueller at

Telephone: (503) 725-3806
Facsimile: (503) 725-3807
Email: donm@eas.pdx.edu

Comprehensive Coverage on JPEG-2000,
MPEG -4 and -7, and H.26x Standards!

With Latest Technology Demos

Seats limited. Please register today!



Date: Fri, 24 Mar 2000 17:35:45 -0500
From: Gerald Kaiser <kaiser@wavelets.com>
Subject: #15 Course: Wavelets, Image Analysis, and Lossless Coding

COURSE TITLE:
A Detailed Introduction to Wavelets
With Applications to Image Analysis and Lossless Coding 

INSTRUCTOR:
Gerald Kaiser
The Virginia Center for Signals and Waves

TIME AND PLACE:
November 13-16, 2000 in Richmond, VA, USA.

This course will be based on an extensive set of recently updated 
lecture notes with MATLAB examples, and include the Instructor's book
"A Friendly Guide to Wavelets."

Enrollment is limited to 10.

For detailed contents and registration information, please visit
http://www.wavelets.com



Date: Thu, 23 Mar 2000 17:44:15 +0100
From: Michael Unser <michael.unser@epfl.ch>
Subject: #16 Job: PhD in Switzerland: Wavelets and splines in medical imaging

Ph.D. position in Switzerland: Splines and wavelets in medical imaging

An assistant position for a Ph.D. candidate is available within the
Biomedical Imaging Group, Swiss Federal Institute of Technology,
Lausanne, Switzerland. The research involves the development of new
spline and wavelet-based algorithms for the processing of biomedical
images; in particular, the detection of cardiac boundaries in
ultrasound images, and their use in 3D/time motion analysis.

The study requirements are a Masters degree in electrical engineering
or applied mathematics. Previous experience in image/signal processing
and wavelets is highly desirable. Applicants should have a strong
theoretical background and a desire to use mathematical tools in their
research. A good knowledge of C, IDL, Matlab, and Mathematica is also
required.

The initial appointment is for one year, renewable in case of
admission to the Ph.D. program. Applications (with detailed CV,
summary of grades, and three reference names) should be sent to

Prof. Michael Unser
Biomedical Imaging Group/IOA/DMT
Swiss Federal Institute of Technology Lausanne
CH-1015 Lausanne EPFL
Switzerland
michael.unser@epfl.ch
http://bigwww.epfl.ch/
E-Mail: michael.unser@epfl.ch
URL    : http://bigwww.epfl.ch/

Tel.: +41(21)693.51.75
Secr.: +41(21)693.51.85
Fax : +41(21)693.37.01



Date: Tue, 28 Mar 2000 08:31:25 +0100 (BST)
From: j.winkler@dcs.shef.ac.uk (Joab Winkler)
Subject: #17 Job: Research Studentships at Sheffield University, UK

PhD studentships are available in The Department of 
Computer Science at The University of Sheffield in 
the following areas :

- Wavelets
- Pattern recognition
- Machine Learning
- Medical decision making
- Non-stationary signal processing

Additional information on the department is available at
http://www.dcs.shef.ac.uk/.

EPSRC studentships are available to eligible students.

The starting date is October 2000, or earlier

Please contact Joab Winkler for more information.

Dr Joab R Winkler
The University of Sheffield
Department of Computer Science
Regent Court
211 Portobello Street
Sheffield S1 4DP
United Kingdom

E-mail : j.winkler@dcs.shef.ac.uk
Tel : +44 (0) 114 222 1834
Fax : +44 (0) 114 222 1810


Date: Wed, 05 Apr 2000 21:26:17 +0200
From: Leif Kobbelt <kobbelt@mpi-sb.mpg.de>
Subject: #18 Cfp: Special issue on subdivision surfaces (CAGD)

                        Call for papers

                Computer Aided Geometric Design

                       (Elsevier Journal)

             Special Issue on Subdivision Surfaces

Subdivision algorithms for the generation of freeform curves and
surfaces have become a mature technique in CAGD and Computer Graphics.
Especially in surface design, subdivision schemes generalize the
concept of shape modeling by control vertices with local influence to
arbitrary control meshes. This allows the designer to build and modify
a globally smooth surface with arbitrary topology as one single piece
with no internal compatibility conditions to be maintained.

After the first subdivision schemes have been proposed in the late
70s, the approach soon gained popularity due to its high efficiency,
simple algorithmic structure, and quality of the generated surfaces.
In contrast, it took quite some time before the theoretic foundations
for the analysis had been developed.

Driven by new applications with increasing requirements, subdivision
algorithms are still a hot topic of current research. For example,
sophisticated smoothing techniques are sought to further enhance
fairness, and interactive multi-scale design techniques are developed
to improve manageability in CAD/CAM systems.  Theoretical
investigations concentrate on further refinement and extensions of
existing results, and on the derivation of special subdivision schemes
with prescribed properties.

To provide a platform for the presentation of recent progress in the
field, CAGD is planning a special issue on the theory and practise of
subdivision algorithms for surfaces of arbitrary topology. In
particular, the following topics are of interest:

  - Theoretical foundations
  - Multiresolution analysis and editing
  - Simulation of physical problems (e.g. FE applications)
  - Interpolation and approximation
  - New schemes with special properties
  - Implementation issues (e.g. data structures, fast evaluation)
  - CAD/CAM applications (e.g. intersection, conversion)

We invite researchers from various fields including geometry,
approximation theory, numerical analysis, and computer graphics to
submit original work in triplicate by

                           June 30, 2000

to one of the two guest editors:

     Ulrich REIF
     TU Darmstadt
     Schlossgartenstra{\ss}e 7
     64289 Darmstadt
     Germany

     Leif KOBBELT
     MPI for Computer Sciences
     Im Stadtwald
     66123 Saarbr{\"u}cken
     Germany



Date: Mon, 10 Apr 2000 18:20:19 -0400 (EDT)
From: Thomas Hogan <hogan@math.ohio-state.edu>
Subject: #19 Contents: J. Approx. Th. - apr00 TOC

Table of Contents: J. Approx. Theory, Volume 103, Number 2, April 2000

D. P. Dryanov and V. Vatchev 
On a polynomial inequality of Paul Erd\H{o}s 
183--205

Pradipta Bandyopadhyay and T. S. S. R. K. Rao 
Central subspaces of Banach spaces 
206--222

Grzegorz W. Wasilkowski and Henryk Wo\'zniakowski 
Complexity of weighted approximation over $\RR^1$
223--251

D. L. Ragozin and J. Levesley 
The density of translates of zonal kernels on compact homogeneous
   spaces 
252--268

Michael Revers 
The divergence of Lagrange interpolation for $|x|^\alpha$ at
   equidistant nodes 
269--280

Ying Guang Shi 
On Christoffel type functions for $L_m$ extremal polynomials, I 
281--291

Salvador Romaguera and Manuel Sanchis 
Semi-Lipschitz functions and best approximation in quasi-metric spaces 
292--301

Ingo Steinwart 
Entropy of $C(K)$-valued operators 
302--328

Dietrich Braess and Christoph Schwab 
Approximation on simplices with respect to weighted Sobolev norms 
329--337

Peter G. Casazza and Ole Christensen 
Approximation of the inverse frame operator and applications to
   Gabor frames 
338--356

Note

J. Szabados 
Where are the nodes of ``good'' interpolation polynomials on the
   real line? 
357--359

Author index for Volume 103 
360



Date: Tue, 21 Mar 2000 08:34:20 -0800
From: Andrew Watson <abwatson@mail.arc.nasa.gov>
Subject: #20 Question: Nasa comparison of wavelet based video codecs

As part of a large, multi-lab study of video quality, we are seeking a
wavelet-based video codec capable of processing full digital TV
resolution (CCIR 601 50/60 Hz) digital color video. We will not be
testing the efficiency of codecs, and the brand names of codecs will
not be reveealed. We want a wavelet codec so that errors different
from those generated by DCT-based codecs can be studied. The codec can
be hardware or software, academic or commercial. Thanks for your help.

Andrew B. Watson
MS 262-2
NASA Ames Research Center
Moffett Field, CA 94035-1000
(650) 604-5419	(650) 604-0255 fax
abwatson@mail.arc.nasa.gov	http://vision.arc.nasa.gov/



Date: Tue, 21 Mar 2000 10:54:33 -0800
From: Bodo Parady <bodo@eng.sun.com>
Subject: #21 Question: Performance of Wavelet based applications

This is for those interested in improving the performance
of wavelet applications by getting computer vendors to
examine wavelet algorithms.

The High Performance Group of SPEC (Standard Performance Evaluation
Corporation www.spec.org) is soliciting OpenMP based applications for
a new benchmark suite.  Applications suitable for the suite must have
more than 90% parallel content and be scalable to 32 cpus or more, and
employ only OpenMP directives to direct parallelism.  Input sets
should run about1 day or more on current one cpu systems. It may be
that wavelet applications are generally too small for consideration.

Become a member of SPEC and participate, or send benchmarks
directly to me or any SPEC HPG member to have it considered
for inclusion in the new suite.

    Bodo Parady
    Technical Representative to SPEC
    Sun Microsystems
    bodo.parady@eng.sun.com



Date: Wed, 22 Mar 2000 17:32:29 -0800 (PST)
From: "T. Mike Chin" <tmc@jpl.nasa.gov>
Subject: #22 Question: Spherical wavelets

Hello.  We are wondering if there is any numerical product
out there for multi-resolution analysis over a sphere (2-D).
We'd like to slice and dice and then put them back altogether
measurements over the Earth.  We've come across with works by 
Gottelman, Sweldens, and others perhaps, and are wondering 
if some of these ideas have been implemented into, say, a
list of coefficients that are matlab-ready.  Am I realistic?
Thanks for your attention,
  Mike Chin
  mike.chin@jpl.nasa.gov



Date: Thu, 23 Mar 2000 12:04:11 +0000
From: Fangmin Shi <fangmin.shi@newport.ac.uk>
Subject: #23 Question: Wavelets and Stereo vision

Dear Sir/Madam,

Can you put my message to the next Wavelet Digest? Thank you.

Hi, I am new to Wavelets. I am looking at the application of wavelets to 
stereo vision (matching). Can anybody provide me any web sites containing 
such document or software for case studies and practice? I especially 
appreciate those who can give me some introductory guide and share their 
personal experience with me. Thanks in advance.

Fran
Fangmin Shi
Ph.D Research,
Mechatronics Research Centre
University of Wales College, Newport
Allt-yr-yn Campus
PO Box 180,
Newport NP20 5XR
South Wales, UK

Tel: +44 01633 432487
Fax: +44 01633 432430



Date: Sun, 26 Mar 2000 23:27:25 +0200
From: "Andrzej Mikulski" <mikulski@uci.agh.edu.pl>
Subject: #24 Question: Wavelets and impulse-type signals

Dear Waveleters,

I am a PhD student at the University of Mining and Metallurgy in Cracow
(Poland) and an application of wavelets in a signal processing of
impulse-type signals is my work. I am looking for references and
articles (pdf, word, ps format) related with wavelet impulse-signal
analysis. I'd like to take advantage of CWT in magnetic tests of steel
ropes, when such signals are the symptom of damage.Is somebody who
interested similar problems?

I'm looking the answer: how minimum fast must be sampled the
impulse-type signal (by means of the A/D converter) that to receive
right results of CWT? Is the Theorem about it?

With many thanks,=20

Andrzej Mikulski

Department of Mechanics and Vibroacoustics,=20

Academy of Mining and Metallurgy

Cracow, Poland

E-mail: mikulski@uci.agh.edu.pl

[Answer WD 9.4 #16]



Date: Mon, 27 Mar 2000 21:36:51 -0800 (PST)
From: Shiva Murthi <shivamurthi@yahoo.com>
Subject: #25 Question: Wavelets for classification

Hi all,

I'm involved in a project that involves using the wavelet transform
for feature extraction of cytometric data for real-time classification
and sorting.I'm currently having limited success with the wavelets and
I was wondering if any of you could help me out by answering the
following:

1.Which basis is ideally suited for feature extraction
problems(PR is not really important)? The signals to
be classified are 1-D pulses representing light
scattered off biological cells(Gaussian shaped pulses
with transient hi-frequency details).

2. In order ro ensure a fairly un-complicated
classifier,typically how many features should be
extracted from the wavelet coefficients and how should
they be extracted?

3. finally, is there any emperical rule of thumb that
states the "ideal" number of levels of wavelet
decomposition for a particular problem?

I will be exxxxtremely grateful if ANYONE can get back
to me with these answers...

Peace out ...Shiva ( shivamurthi@yahoo.com )



Date: Tue, 28 Mar 2000 18:12:56 +0800
From: Fabio Boschetti <fabio@ned.dem.csiro.au>
Subject: #26 Question: Estimation of Lipschitz exponent 

Hello,

I am interested in receiving references to papers and work on the 
estimation of Lipschitz coefficients via wavelet analysis and/or multiscale 
edges representation.

Also, practical experience (applications / did it work well? / if not, why 
not? / robustness) would be most welcome.

Thanks in advance

Fabio
Dr. Fabio Boschetti

Senior Research Scientist
CSIRO - Division of Exploration and Mining
39 Fairway, Nedlands, 6009, Western Australia
Tel ++61 8 92848421
Fax ++61 8 93891906
E-mail fabio@ned.dem.csiro.au
Web page http://www.ned.dem.csiro.au/BoschettiFabio



Date: Mon, 17 Apr 2000 11:12:01 -0700 (PDT)
From: Irislav Drezga <idrezga@yahoo.com>
Subject: #27 Question: Music Recognition & wavelets

Hi there

I am currently working on recognition of songs.  I did some research
and found wavelets mentioned in this contex.  The idea is, as far as I
understand it, to calculate (discrete) wavelet coefficients for songs
and then compare them with a candidate song.  >From my understanding
of the problem, at least it is needed to calculate time-frequency
relationship for each song.  However, both Discrete Wavelet Transform
and Windowed Fourier Transform can produce time-frequency
relationship.

Since I am new to wavelets I would appreciate very
much answers to the following questions:

1) Between wavelets and Fourier approach, which
approach is to recommend in the context of music
recognition?

2) Are wavelet coefficients faster to compute than
Fourier in the case of Windowed Fourier Transfrom?

3) Is there any particular wavelet "mother function"
to use when dealing with auditory signals (note: I  do
not need to reproduce the song, though sparse
representation would be desirable).

4) Is there any research or papers or available
computer code that you can recommend in pursuing the
problem.

Thank you very much in advance,
Irislav Drezga.


<webmaster@wavelet.org>