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

The Wavelet Digest, Volume 8, Issue 4 (April 15, 1999)




Today's Editor: Wim Sweldens
                Lucent Technologies, Bell Laboratories
                wim@bell-labs.com



Today's Topics:

     1: Book:     Spline Functions and the Theory of Wavelets
     2: Preprint: Covariance structure of wavelet coefficients
     3: Preprint: Adaptive Wavelet Analysis
     4: Preprint: Wavelet Analysis: Applications in Industry
     5: Preprint: 2 papers on approximation in wavelet-like spaces
     6: Preprint: Wavelets on irregular point sets
     7: Preprint: Two preprints from Karsten Urban
     8: Preprint: The relation of the Allan- and Delta-variance...
     9: Preprint: HB-splines, a Blend of Hermite splines and B-splines
    10: Preprint: Smoothing irregularly sampled signals
    11: Thesis:   Ph.D. Thesis on Sampling
    12: Thesis:   Wavelet Transforms and Image and Video Compression
    13: Meeting:  Internat. Conf.On Wavelet Analysis And Its Applications
    14: Job:      Student/Summer Intern Positions in Malibu
    15: Answer:   Looking for 2-D wavelet Fortran code (WD 8.3 #14)
    16: Answer:   Adaptive construction of mother wavelets (WD 8.3 #15)
    17: Question: Wavelets vs. FFT
    18: Question: Wavelets and Eigenvalues
    19: Question: Wavelet Features Extraction
    20: Question: Denoising Signals/Images with Non-Constant Noise Variance
    21: Question: How to order a book fron P.R.C.
    22: Question: Wavelets and denoising for high voltage transforms
     
Current number of members: 13608

E-mail:
   General help:                help@wavelet.org
   Add yourself as a member:    add@wavelet.org
   Remove yourself as a member: remove@wavelet.org
   Publish in the next WD:      publish@wavelet.org
   Receive publishing policy:   policy@wavelet.org

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

Calendar of events:

 May 31-Jun 12: Summer School on Multiscale PDE, Marseille, France
 Jun 14-19: Mathtools 99, St-Petersburg, Russia
 Jun 14-16: Fractals in Engineering, Delft, The Netherlands
 Jun 16-18: Wavelet workshop by Gilbert Strang, Braunschweig
 Jul 1-7:   Curves and Surfaces, St-Malo, France
 Jul 4-8:   Wavelet Conference, 3rd IMACS/IEEE, Athens, Greece
 Jul 6-16:  Mathematical Aspects of Image Processing
 Jul 18-23: SPIE Wavelet Applications VII, Denver, CO
*Nov 15-20: Int. Conf. on Wavelet Analysis and its Applications

--- 2000
 May 17-20: Trends in Approximation Theory 2000, Nashville, TN
 Mar 12-16: Australian Society for Exploration Geophysics, Perth








Date: Mon, 12 Apr 1999 15:21:00 -0400
From: Gilles.Deslauriers@mail.polymtl.ca (Gilles Deslauriers)
Subject: #1 Book: Spline Functions and the Theory of Wavelets

Spline Functions and the Theory of Wavelets, 1999
Serge Dubuc, Gilles Deslauriers, editors.

CRM (Centre de Recherches Mathématiques, Université de
Montréal) Proceedings & 
Lectures Notes   (ISSN 1065-8580; v.18)
American Mathematical Society
ISBN 0-8218-0875-3

Contents

Preface

    Slines functions emerged from a 1946 article of Schoenberg and owe
much of their development to de Boor while the theory of wavelets was
born in the work of Morley and Grossmann in 1980 and took off as a
results of a number of authors among which Meyer and Daubechies are
the most reknowned.

    Among the historical bench marks in the development of the theory
of wavelets is the discovery by Meyer of the first orthogonal wavelet
basis. This occured in 1985 in France, in a manner that Meyer himself
described as mysterious and accidental.  Shortly thereafter Mallat
introduced multiresolutional analysis which explained some of the
mysteries of Meyer's wavelet.  Then on a cold, snowy February day in
1987 in Montréal, Daubechies found a wavelet with compact
support. Next came biorthogonal wavelets which were related to spline
functions. In Paris in May 1985, Deslauriers and Dubuc used dyadic
interpolation to explain multiresolution, a term which by then was
familiar to many.  Since these developments, alot of water which has
gone under the bridge (much of it French) and we might also say that
many waves and wavelets have also gone under that bridge. Within a few
years, a multitude of books and other publications on the theory of
wavelets appeared, several conferences were held, and congresses were
organized.  Was this the birth of a theory ? During this same period,
related subjets, such as multiresolutional analysis, subdivision
schemes, dilatation equations, to cite just a few, were developped.
As a result of this brew of ideas there were applications to physics,
signal processing and imaging, the theory of fractals, statistics, and
computer science. These interactions allowed us to clarify the
theory. Mainwhile the development of the theory of splines proceeded
in a parallel fashion.

    From all this, emerged the idea of holding a series of thematic
workshops relating the theory of wavelets and the theory of splines
whitout losing sight of related applications which were of increasing
importance. These workshops took place at the Centre de recherches
mathématiques in 1996. Ten themes were treated, each for one week :
Geometric Modelling with Splines, Splines in Approximation and in
Differential Equations, Splines and Wavelets, Wavelets and
Approximations, Multiresolution Analysis and Subdivision Operators,
Wavelets and Differential Equations, Wavelets in Signal Processing and
Image Analysis, Wavelets and Fractals, Wavelets in Physics and finally
Splines and Wavelets in Statistics. The contributions of Professor
Yves Meyer who gave a series of lectures as Aisenstadt chaiholder,
were most appreciated.  They resulted in a book entitled Wavelets,
Vibrations an Scalings published in the CRM Monograh Series (Volume 9)
by the American Mathematical Society.  The proceeding of these
workshop consist of 27 contributions which were grouped in four parts
entitled Spline Functions, Theory of Wavelets, Wavelets in Physics,
Splines and Wavelets in Statistics. Each section begins whith a
presentation due respectively to Brunner, Jaffard, Arnéodo, and
MacGibbon.

Gilles Deslauriers



Date: Mon, 29 Mar 1999 16:22:43 -0600
From: "Marina Vannucci" <mvannucci@stat.tamu.edu>
Subject: #2 Preprint: Covariance structure of wavelet coefficients

Title: Covariance Structure of Wavelet Coefficients:
       Theory and Models in a Bayesian Perspective.

Authors: Vannucci, Marina (Texas A&M University, USA)
         Corradi, Fabio (University of Florence, Italy)

Abstract:

In this paper we present theoretical results on the random wavelet
coefficients covariance structure. We use simple properties of the
coefficients to derive a recursive way to compute the within- and
across-scale covariances. We point out a useful link between the
proposed algorithm and the two-dimensional discrete wavelet transform.
We then focus on Bayesian wavelet shrinkage for estimating a function
from noisy data. A prior distribution is imposed on the coefficients
of the unknown function. We show how our findings on the covariance
structure make it possible to specify priors that take into account
the full correlation among coefficients through a parsimonious number
of hyperparameters. We use Markov chain Monte Carlo methods to
estimate the parameters and illustrate our method on benchmark
simulated signals.

Status: to appear in the Journal of the Royal Statistical Society, Series B

Postscript file can be found at

http://stat.tamu.edu/~mvannucci/webpages/CV.html

Marina Vannucci                       
Department of Statistics                  406D  Blocker Building
Texas A&M University                     Phone:  +1-409-845-0805
College Station                          Fax:    +1-409-845-3144
TX 77843-3143              URL: http://stat.tamu.edu/~mvannucci/



Date: Fri, 02 Apr 99 12:36:08 -0500
From: "L.Novikov" <nov@iai.rssi.ru>
Subject: #3 Preprint: Adaptive Wavelet Analysis

Title: Adaptive Wavelet Analysis of Signals

Abstract:The relationship between matched filtering and wavelet
analysis is discussed. A unique wavelet synthesis procedure is offered
which takes into account a priori information on the signal shape
(apparatus function) and noise correlation function. This makes
possible adaptation of the wavelet analysis of measurement data for
best signal denoising and baseline drift minimization. The results of
modeling based on the Gaussian signal model are given.

The full text of this paper can be found at the web-site

http://www.iai.rssi.ru/novik/adaptDWT_ps.zip



Date: Thu, 8 Apr 1999 10:38:21 +0900
From: MEI@jp.ibm.com
Subject: #4 Preprint: Wavelet Analysis: Applications in Industry

Preprint available:
Mei Kobayashi
Wavelet Analysis: Applications in Industry
IBM Tokyo Research Laboratory Research Report RT-0314

Abstract: In this paper, we describe how wavelet transforms can be
used for analysis of data, signals, and mathematical models and how
this analysis has been employed in system prototypes, commercial
products and industrial design and manufacturing processes.

Key words: multiresolution analysis, time-frequency
analysis, wavelet transforms, wavelets

This report has an extensive list of wavelet-related patents
filed in the United States between Jan. 1991 and March 31, 1999.

or contact: Mei Kobayashi
            IBM Research, Tokyo Research Laboratory
            1623-14 Shimotsuruma, Yamato-shi, Kanagawa-ken 242-8502
            Japan
            Mei_Kobayashi@jp.ibm.com



Date: Tue, 13 Apr 1999 13:01:25 +0200
From: Thierry BLU <thierry.blu@epfl.ch>
Subject: #5 Preprint: 2 papers on approximation in wavelet-like spaces

The following papers have recently been accepted for publication in
the IEEE Transactions on Signal Processing. They both deal with the
evaluation of approximation techniques, the second one focusing on
wavelets.

                         TITLE
  Quantitative Fourier Analysis of Approximation Techniques:
        Part  I - Interpolators and Projectors.
        Part II - Wavelets.

                        AUTHORS
            Thierry BLU and Michael UNSER
   (Swiss Federal Institute of Technology, Lausanne)

                        WEB LINK
     http://bigwww.epfl.ch/publications/blu9901.html
     http://bigwww.epfl.ch/publications/blu9902.html

                     ABSTRACT Part I We present a general
Fourier-based method that provides an accurate prediction of the
approximation error as a function of the sampling step T.  Our
formalism applies to an extended class of convolution-based signal
approximation techniques, which includes interpolation, generalized
sampling with prefiltering, and the projectors encountered in wavelet
theory.  We claim that we can predict the L2-approximation error, by
integrating the spectrum of the function to approximate---not
necessarily bandlimited---against a frequency kernel E(w) that
characterizes the approximation operator.  This prediction is easier,
yet more precise than was previously available.  Our approach has the
remarkable property of providing a global error estimate that is the
average of the true approximation error over all possible shifts of
the input function.  Our error prediction is exact for stationary
processes, as well as for bandlimited signals.  We apply this method
to the comparison of standard interpolation and approximation
techniques.  Our method has interesting implications for approximation
theory.  In particular, we use our results to obtain some new
asymptotic expansions of the error as T tends to 0, and also to derive
improved upper bounds of the kind found in the Strang-Fix theory.  We
finally show how we can design quasi-interpolators that are
near-optimal in the least-squares sense.

                     ABSTRACT Part II In a previous paper, we proposed
a general Fourier method which provides an accurate prediction of the
approximation error, irrespective of the scaling properties of the
approximating functions.  Here, we apply our results when these
functions satisfy the usual two-scale relation encountered in dyadic
multiresolution analysis.  As a consequence of this additional
constraint, the quantities introduced in our previous paper can be
computed explicitly as a function of the refinement filter.  This is
in particular true for the asymptotic expansion of the approximation
error for biorthonormal wavelets, as the scale tends to zero.  One of
the contributions of this paper is the computation of sharp,
asymptotically optimal upper bounds for the least-squares
approximation error.  Another contribution is the application of these
results to B-splines and Daubechies scaling functions, which yields
explicit asymptotic developments and upper bounds.  Thanks to these
explicit expressions, we can quantify the improvement that can be
obtained by using B-splines instead of Daubechies wavelets.  In other
words, we can use a coarser spline sampling and achieve the same
reconstruction accuracy as Daubechies: Specifically, we show that this
sampling gain converges to pi as the order tends to infinity.



Date: Tue, 13 Apr 1999 16:08:52 -0400
From: Wim Sweldens <wim@bell-labs.com>
Subject: #6 Preprint: Wavelets on irregular point sets

Title: Wavelets on irregular point sets

Authors: Ingrid Daubechies   Princeton Univerisity
         Igor Guskov         Princeton Univerisity
         Peter Schroeder     CalTech
         Wim Sweldens        Bell Labs

Abstract: In this article we review techniques for building and
analyzing wavelets on irregular point sets in one and two
dimensions. We discuss current results both on the practical and
theoretical side. In particular we focus on subdivision schemes and
commutation rules. Several examples are included.

Status: To be published in Phil. Trans. R. Soc. Lon. A. 

/who/wim/papers/royal/



Date: Wed, 14 Apr 1999 10:25:32 +0200
From: "Dr. Karsten Urban" <urban@dragon.ian.pv.cnr.it>
Subject: #7 Preprint: Two preprints from Karsten Urban

The following two preprints are online available:

Title: "On the adaptive computation of integrals of wavelets"
Authors: Silvia Bertoluzza, Claudio Canuto, Karsten Urban
Preprint No. 1129, Istituto di Analisi Numerica del C.N.R. Pavia, 1999. 
URL: ftp://www.igpm.rwth-aachen.de/pub/urban/reports/integrals.ps.gz

Abstract: The numerical solution of partial differential equations
involves the computation of integrals of products of given functions
and (derivatives of) trial and test functions. We study this problem
using adaptively chosen wavelet bases.  Firstly, we reduce this
problem to the computation of 1-dimensional integrals.  Then, we
consider appropriate adaptive approximations and study the induced
error.  Finally, we present an algorithm for computing these integrals
and give numerical results.

Title: "Adaptive Wavelet Methods for Saddle Point Problems"
Authors: Stephan Dahlke, Reinhard Hochmuth, Karsten Urban
Preprint No. 1126, Istituto di Analisi Numerica del C.N.R. Pavia, 1999.
URL: ftp://www.igpm.rwth-aachen.de/pub/urban/reports/saddle.ps.Z

Abstract: Recently, adaptive wavelet strategies for symmetric,
positive definite operators have been introduced that were proven to
converge. This paper is devoted to the generalization to saddle point
problems which are also symmetric, but indefinite.

Firstly, we derive explicit criteria for adaptively refined wavelet
spaces in order to fulfill the {Ladyshenskaja-Babuska-Brezzi (LBB)}
condition and to be fully equilibrated.  Then, we investigate a
posteriori error estimates and generalize the known adaptive wavelet
strategy to saddle point problems. The convergence of this strategy
for elliptic operators essentially relies on the positive definite
character of the operator.  As an alternative, we introduce an
adaptive variant of Uzawa's algorithm and prove its convergence.

Finally, we detail our results for two concrete examples of saddle
point problems, namely the mixed formulation of the Stokes problem and
second order elliptic boundary value problems where the boundary
conditions are appended by Lagrange multipliers.

Karsten Urban
Istituto di Analisi Numerica del C.N.R.
via Abbiategrasso, 209, I-27100 Pavia, Italy
urban@dragon.ian.pv.cnr.it



Date: Thu, 15 Apr 1999 14:36:20 +0200
From: Maik Zielinsky <zielinsky@ph1.uni-koeln.de>
Subject: #8 Preprint: The relation of the Allan- and Delta-variance...

Title: The relation of the Allan- and Delta-variance to the continuous
wavelet transform

Authors: M. Zielinsky, J. Stutzki

Abstract: This paper is understood as a supplement to the paper "On
the fractal structure of molecular clouds" by [Stutzki et al, 1998],
where we have shown the usefulness of the Allan-variance and its
higher dimensional generalization, the Delta-variance, for the
characterization of molecular cloud structures. In this study we
present the connection between the Allan- and Delta-variance and a
more popular structure analysis tool: the wavelet transform. We show
that the Allan- and Delta-variances are the variances of wavelet
transform coefficients.

Status:  Accepted by Astronomy and Astrophysics

Available on the Web at:
   http://xxx.lanl.gov/abs/astro-ph/9904190



Date: Thu, 15 Apr 1999 13:52:42 +0200 (MEST)
From: grisoni@labri.u-bordeaux.fr ( Laurent Grisoni)
Subject: #9 Preprint: HB-splines, a Blend of Hermite splines and B-splines

Dear waveleters,
        
The preprint:

"HB-splines: a Blend of Hermite splines and B-splines"

is available at
http://dept-info.labri.u-bordeaux.fr/~grisoni/

and has been accepted for publication in Computer Graphics Forum.
Here is the abtract of this paper:

This paper proposes to study a spline model, called HB-splines, that
is in fact a B-spline representation of Hermite splines, combined with
some restriction on the differential values at segment
boundaries. Although this model does not appear able to offer
something new to the computer graphics community, we think that
HB-splines deserve to be considered for themselves because they embed
many interesting features. First, they include all the classical
properties required in a geometric modeling environment (convex hull,
local control, arbitrary orders of parametric or geometric
continuity). Second, they have a nice aptitude for direct manipulation
(i.e. manipulation without using control points).  For this purpose,
we propose a new graphic widget, called control sails, that offers the
user an intuitive way to specify local properties (position, tangent,
curvature) of a curve or a surface. Finally, they provide an elegant
formulation of a biorthogonal wavelet family, that permits
multiresolution manipulations of the resulting curves or surfaces, in
a very efficient way.

keywords: geometric modeling, splines, direct manipulation, 
multiresolution.

Authors:
Laurent Grisoni, Carole Blanc, Christophe Schlick
LaBRI, Universite Bordeaux I,
351 cours de la Liberation,
33405 Talence,
FRANCE.

Any comments, questions, or remarks are of course welcome. 



Date: Thu, 15 Apr 1999 14:25:06 +0200 (MET DST)
From: Maarten Jansen <Maarten.Jansen@cs.kuleuven.ac.be>
Subject: #10 Preprint: Smoothing irregularly sampled signals

Title: Smoothing irregularly sampled signals using wavelets and cross
       validation 

Authors: Maarten Jansen and Adhemar Bultheel, K.U.Leuven, Belgium

Abstract: 

Coefficient thresholding is a popular method in wavelet based noise
reduction. A wavelet decomposition is typically a sparse
representation of noise-free signals: the essential information is
captured by a limited number of large, important coefficients, while
the main part of coefficients is close to zero. Replacing these small
coefficients by zero is a straightforward way to reduce noise variance
without affecting the noise-free signal too much.  Recently,
algorithms have been developed for wavelet decompositions of
non-equidistant samples, using the so called lifting scheme and second
generation wavelets. We investigate how to apply these algorithms to
reduce noise in signals on a non-equidistant grid.  The paper also
illustrates that the method of generalized cross validation in these
settings still succeeds in finding good thresholds.  Unlike other
methods, we do not `precondition' the input, but use the lifting
scheme to deal with the irregularity of the grid. This approach seems
more natural and shows promising results. However, instability
problems arise from the actual scheme. This article describes the
method and explains in what cases problems may occur.

Status: Technical Report, Department of Computer Science

Postscript file can be found at:
http://www.cs.kuleuven.ac.be/~maarten/publications/noneqTW.ps.gz

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



Date: Thu, 8 Apr 1999 10:39:36 +0900
From: MEI@jp.ibm.com
Subject: #11 Thesis: Ph.D. Thesis on Sampling

Ph.D. Thesis by Wen Chen 1999
Title: Sampling for Non-bandlimited Signals
Thesis Advisor: Shuichi Itoh
Institution: Graduate School of Information Systems
             University of Electro-Communications
             Chofugaoka 1-5-1, Chofu, Tokyo 182 Japan
             fax: 81+424-42-8092



Date: Tue, 13 Apr 1999 15:33:50 +0100
From: =?iso-8859-1?Q?=D8yvind=20Str=F8mme?= <os@cs.strath.ac.uk>
Subject: #12 Thesis: Wavelet Transforms and Image and Video Compression

PhD Thesis : On the Applicability of Wavelet Transforms to Image and
              Video Compression
Author : �Øyvind Str�ømme
Date : 29 March 1999
Institution : University of Strathclyde, Glasgow, UK
Supervisor : Prof. Douglas R McGregor

Committee:

Prof. Roger J Clarke - External examiner
Prof. Andrew D McGettrick - Chairman
Dr. Richard J Fryer - Internal examiner

Abstract:

Image and Video compression techniques aim for a perceptually faithful
representation of the original image for given constraints of
compression ratio and processing power. This thesis critically reviews
current approaches, in particular those using the Wavelet concept. A
key stage in image compression is detail recognition and efficient
representation of regularly patterned texture. In naturalistic images
such patterns are coherent over limited image areas. The Wavelet model
is particularly well suited to capture limited coherence. There are,
however, many candidate decomposition methods, Wavelet functions, and
final encoding techniques.

A novel technique for speeding up Wavelet transformations is
presented.  Further, novel results show a little known image
decomposition as more economical than the conventional technique. New
results include a co-encoding method improving lossless compression,
and an evaluation of Wavelet filters indicating a possible best
performer.

An extensive experimental investigation is reported which utilised an
internationally accepted benchmark of images. Peak Signal to Noise
Ratio (PSNR) was used as objective measure of reproduction quality for
comparison of the different approaches.

An investigation is reported of a number of Wavelet-associated methods
aiming to determine optimal quality and minimal computational
resources.  Both theory and experiments are pursued.

A novel hybrid Wavelet technique is described reducing computational
requirements by a factor of 3, while providing equivalent quality as
compared to existing Wavelet methods. 

The Quincunx decomposition is reported most cost-efficient, providing
equivalent quality to the conventional Pyramidal decomposition, though
the latter requires over 25% more computational resources.

A novel coefficient-encoding method is described providing a 7-8%
improvement over existing approaches. 

Comprehensive experimental comparisons of Wavelet functions show that
the biorthogonal 9/7 Wavelet performs best of the commonly used Wavelet
functions. 

The thesis provides comprehensive results on Wavelet techniques as
applied to image and video compression, which will hopefully prove
useful to designers of compression systems.

If you want a copy, send me an email or write to

�Øyvind Str�ømme
Department of Computer Science
University of Strathclyde
Livingstone Tower
26 Richmond street
Glasgow G1 1XH
UK
tel.: +44-141-552-4310
fax: +44-141-552-5330
email: oyvind@i.am
web: i.am/oyvind



Date: Wed, 14 Apr 1999 14:39:58 -0500
From: "MTH_JXW" <MTH_JXW@shsu.edu>
Subject: #13 Meeting: Internat. Conf.On Wavelet Analysis And Its Applications

International Conference On Wavelet Analysis And Its Applications

                 November 15 -- November 20, 1999
           Zhongshan University, Guangzhou, P. R. China

An International Conference on Wavelet Analysis and its Applications
will be held in Guangzhou between November 15 and November 20, 1999.
The objective of this conference is to bring together researchers in
wavelet analysis and related areas to discuss new developments, to
exchange ideas, and to collaborate in research.

ADDRESS:

      Department of Mathematics
      Zhongshan University
      Guangzhou  510275
      P. R. China

SCOPE AND TOPICS:

I. Wavelet Theory:

Constructions and properties of wavelet bases
Wavelet frames
Wavelets in Harmonic Analysis
Fractal functions
Wavelet approximation and numerical analysis

II. Wavelet Applications:

Wavelets and filter banks
Image and video compression
Signal and speech processing
Computer vision
Biomedicine image processing
Remote sensing
Statistical data denoising

SPONSORS:

The Ministry of Education of China
The National Natural Science Foundation of China
Peking University, China
Zhejiang University, China
Zhongshan University, China

SCIENTIFIC COMMITTEE:

Qingyun Shi, Peking University, China (Chairperson)

Charles K. Chui, Texas A&M University, USA (Chairperson)

Donggao Deng, Zhongshan University, China
Daren Huang, Zhejiang University, China
Lizhong Peng, Peking University, China

Rong-Qing Jia, University of Alberta, Canada
Iain Johnstone, Stanford University, USA
Jianzhong Wang, Sam Houston State University, USA
Guido Weiss, Washington University, USA

ORGANIZING COMMITTEE:

Donggao Deng, Zhongshan University, China
Daren Huang, Zhejiang University, China
Wei Lin, Zhongshan University, China
Lizhong Peng, Peking University, China
Xingwei Zhou, Nankai University, China

Rong-Qing Jia, University of Alberta, Canada
Jianzhong Wang, Sam Houston State University, USA

PRINCIPAL INVITED SPEAKERS:

      A. Aldroubi (Vanderbilt University, USA)
      Donggao Deng (Zhongshan University, China)
      D. L. Donoho (Stanford University, USA)
      T. N. T. Goodman (University of Dundee, UK)
      D. Hardin (Vanderbilt University, USA)
      P. Jorgensen (University of Iowa, USA)
      K. S. Lau (Chinese University of Hong Kong)
      S. L. Lee (National University of Singapore)
      C. A. Micchelli (IBM Research Center, USA)
      S. Riemenschneider (University of Alberta, Canada)
      Qingyun Shi (Peking University, China)
      J. Ward (Texas A&M University, USA)
      G. L. Weiss (Washington University, USA)
      V. Wickerhauser (Washington University, USA)
      Yuesheng, Xu (University of North Dakota, USA)



Date: Sat, 27 Mar 1999 15:07:55 -0800 (PST)
From: Trish Keaton <keaton@isl.hrl.hac.com>
Subject: #14 Job: Student/Summer Intern Positions in Malibu

Intern Positions at HRL in Malibu, CA

The Signal and Image Processing Department at HRL Laboratories
(formerly Hughes Research Laboratories) located in Malibu, CA
http://www.hrl.com has intern positions for summer/fall 1999.  Our
current projects include audio/video surveillance, computer vision
techniques for automotive safety and human computer interaction,
robust speech recognition, decision aids and information fusion
methods for image exploitation.

Candidates should be Ph.D. students pursuing degrees in either
Electrical Engineering or Computer Science with an emphasis in at
least one of the following areas: signal, speech, and image
processing, computer vision, pattern recognition, Bayesian networks,
and other information fusion methods. Candidates should be self
motivated, have very strong communication skills, and possess
programming skills in either MATLAB, C, C++, or Java.  Specific duties
will include theoretical analysis, algorithm design, and software
simulation. Based on government restrictions regarding the export of
technical data, a U.S. citizen or permanent resident status is
required for most positions in our organization.

Please send resumes to:

Sheryl Wells
Human Resources
HRL Laboratories
BLDG 250, M/S RL73
3011 Malibu Canyon Road
Malibu, CA  90265
swells@hrl.com

Email correspondence is preferred. Files should be in either
text, or MS Word format and emailed with the subject heading: 
Intern Position - RM2.



Date: Thu, 01 Apr 1999 20:09:06 +0300
From: Isar Alexandru <isar@ee.utt.ro>
Subject: #15 Answer: Looking for 2-D wavelet Fortran code (WD 8.3 #14)

Question:

From: Daniele Colombo <<dcolombo@geosystem.net>
Subject: #14 Question: Looking for 2-D wavelet transform Fortran code

> Hi, I am looking for software or FORTRAN routines to implement
> wavelet transforms in 2-D. The aim is to apply wavelet filters to
> separate anomalies in 2-D distributions of geophysical parameters.
> If somebody is aware of available fortran routines or programs to
> perform this task please let me know.

Answer:

In the abstract of the paper of Umberto Amato and Dan Tudor Vuza with
the title "A Collection of Routines for the Wavelet Transform of
Daubechies Type" we can read:

"Main aim of the present paper is to provide a collection of routines
written in Fortran and C++ language for the evaluation of the wavelet
ransform of Daubechies type. ... Special Fortran code is developed for
2D transform".

Because you are from Italy I suggest to find Umberto Amato: Instituto
per Applicazioni della Matematica CNR, via Pietro Castellino 111,
80131 Napoli.

You can obtain his e-mail addres from his romanian collegue Dan Tudor
Vuza dvuza@stoilow.imar.ro



Date: Thu, 15 Apr 1999 11:38:53 +1000
From: Dennis Lee <dennis.lee@eng.monash.edu.au>
Subject: #16 Answer: Adaptive construction of mother wavelets (WD 8.3 #15)

The following reply to the question for designing a wavelet using
neural network for application in condition monitoring- wavelet digest
vol.8(3)

> The reply shown below was from Dr. Shubha Kadambe
> Information sciences laboratories
> Malibu
>
> In our experience constructing a mother wavelet that is optimum for
> a given data is more effective using neural netwok architecture. If
> you are interested in our work, please refer to two journal papers
> published in Optical Engg special issues on wavelets (Sep. 1992) and
> adaptive wavelets (Jul 1994).  

Thanks for Dr.  Shubha Kadambe and wavelet digest for information.

   regards
   Dennis Lee
   Monash University, Australia
   email: dennis.lee@eng.monash.edu.au



Date: Fri, 26 Mar 1999 11:09:20 +0300
From: "SP Groupe" <spg@fel.ntu-kpi.kiev.ua>
Subject: #17 Question: Wavelets vs. FFT

What pricipal advantage of Wavelet analysis prosesses over the FFT
analysis. They say WA is also good because it manages with
non-stationary signals. Nevertheless the determination of
NON-STATIONARY DETERMINED SIGNAL does not exist If there is
non-stationary RANDOM process than the determination of non-stationary
occational process is known and cleared. But in this case there is the
dignity of WA is limited by the area of usefulness only random
processes! It is likely not so. It would be desirable to get
explanations - which of analytical properties WA ensures this
advantage WA. How this property can be treated - physically or
geometrically?  Would you send either the direct answer or the
references on the publication in Internet which contains the answers
our questions.  SPG



Date: Sat, 27 Mar 1999 13:23:24 +0000
From: Ian Brayshaw <Ian.Brayshaw@dpr.csiro.au>
Subject: #18 Question: Wavelets and Eigenvalues

Hi.

I'm new to the field of wavelets and for my undergraduate thesis in
Engineering am trying to find a method of accelerating the extraction
of eigenvalues (and eventually eigenvectors) from matrices. The
proposal is to use wavelets to accelerate the eigenvalue extraction
for a (possible) subset of all eigenvalues of a given matrix, by
combining it with such eigenvalue extraction techniques as the
subspace iteration method. My first question is has this been done
before?  The matrix systems I will be examining in particular are
those generated from dynamic finite element analysis of vibrating
structures.

One suggestion has been to use the wavelet transform as a
preconditioner to transform the matrices into a form that lends itself
to rapid eigenvalue extraction (ie diagonal, triangular, etc). Another
has been to visualise the iteration processes as analyzing a spectrum
of possible eigenvalues and hence use wavelets to extract information
from the spectrum.

Am I barking up the wrong tree? Will wavelets be a suitable tool? This
is not the typical application of wavelets and literature is somewhat
scarce.

Hope you can assist me in my search.

Thanks.

Ian

PS      As per the Publications Policy, I'll collate any responses and forward 
        them to the Digest.

 Ian Brayshaw   Australian Petroleum Resources CRC  Ian.Brayshaw@dpr.csiro.au
                CSIRO Petroleum                       Tel:    +61 3 9259 6876
     __|__      PO Box 3000, Glen Waverley            Fax:    +61 3 9259 6900
 --o--(_)--o--  VICTORIA    AUSTRALIA    3150         http://www.dpr.csiro.au



Date: Sun, 28 Mar 1999 00:49:04 +0800
From: rich <asterix@tartarus.uwa.edu.au>
Subject: #19 Question: Wavelet Features Extraction

Hi everyone,

I'm new to the digest and even more so, new to the world of wavelets.
So please bear with me :)

I am currently doing a project involving analysis of abnormalies in
ECG - electrocardiogram. I am thinking of using wavelet coeefficients
as features descriptors to feed into my artificial neural network, and
hence train the network to differentiate between the different classes
of abnormalies. I have only just begun reading on wavelets, and as far
as I understood, the wavelet transform gives me two type of
coefficients: approximation and details coefficients (I made the
transform using cwt and dwt from the Matlab wavelet toolbox). Now I am
not quite sure if indeed these are the coefficients I want to be
using, because what I was looking for are the coefficients infront of
the basis functions, such as those infront of sines and cosines in a
Fourier transform.

Any pointers/references are greatly appreciated. Please forward
replies to:

law-kr@ee.uwa.edu.au

Thanks in advance.

Richard Law

[Answer WD 8.5 #19]
[Answer WD 8.5 #20]



Date: Mon, 05 Apr 1999 23:53:47 +0100
From: Brett Keating <brettk1@home.com>
Subject: #20 Question: Denoising Signals/Images with Non-Constant Noise Variance

Hello, I am a PhD student at UCDavis Dept. of Applied Science and I
work at LLNL. I am interested in denoising images that contain a lot
of important detail. The noise in the image is not uniform, i.e. parts
of the image have a higher local noise variance than other parts that
may even have no noise at all. I know where the noise is and how
strong it is, due to the image creation process. The noise variance in
the image can be computed on a per-pixel basis, i.e. the noise
variance is a function of position in the image. I want to preserve as
much detail as possible, thus I only want to remove noise where noise
is present... however, most denoising algorithms I have found assume a
uniform noise variance over the whole image. Are there any approaches
(esp. in the literature) that consider noise as a function of position
in the image? I am particularly comfortable with soft-thresholding, so
if there is a closely related method it would be most helpful.

Thank you,
Brett Keating
brettk1@home.com

[Answer WD 8.5 #21]



Date: Fri, 9 Apr 1999 12:00:28 -0400 
From: "Lin, Tsan" <tsanlin@nvl.army.mil>
Subject: #21 Question: How to order a book fron P.R.C.

I would like to order a copy of the book entitled "Wavelet analysis &
Signal Processing ---Theory, Applications & Software implementation"
by Jiang Ping Li. I tried to contact the author by email but the email
address shown in March Wavelet Digest is not a valid address. I would
appreciate if any one can help me to contact the publisher or author.

Thanks,

Tsan Lin tsanlin@nvl.army.mil



Date: 10 Apr 1999 12:24:19 -0000
From: deq_li@163.net
Subject: #22 Question: Wavelets and denoising for high voltage transforms

Dear waveleteers:

I'm a postgraduate sdudent.I'm working on my master thesis"Patial
Discharge(PD) dectection in high voltage transform substation".I
decide to use wavelet transform for denoising.There's a difficulty in
front me:I donn't know how to descriminate the partial discharge on
the wire open in air(called "corona") from that in substation,their
freqency are almost the same.The difference is that one > in oil one
in air(which is more steady in pulse amplitude .Any one who knows how
to tackle this problem and would like to help me?

Thanks in advance.
My E-mail: deq_li@163.net or d.q.li@263.net

Yours,sincerely! 
Li Dequan
Sience school of Chongqing University
Chongqing,P.R.China (zip400044)

<webmaster@wavelet.org>