wavelet digest, vol 10, nr 04.txt

现代信号处理讲义,包含许多的例题,解答等Qinghua University publishing house modern signal processing course

Note


Akio Arimoto
Completeness of trigonometric system with integer indices
   $\{e^{inx}; x\in\RR\}$
311--317


Book reviews
318--325


Book review reply
326


Table of Contents: J. Approx. Theory, Volume 113, Number 2, December 2001


Guanzhen Zhou and Songping Zhou
Some remarks on equivalence of moduli of smoothness
165--171


Michael J. Johnson
Scattered date interpolation from principal shift-invariant spaces
172--188


Jes\'{u}s de la Cal and Javier C\'{a}rcamo
On certain best constants for Bernstein-type operators
189--206


Yang Wang
Subdivision schemes and refinement equations with nonnegative masks
207--220


Holly Carley, Xin Li, and R N. Mohapatra
A sharp inequality of Markov type for polynomials associated with
   Laguerre weight
221--228


M. P. H. Wolff
Discrete approximation of unbounded operators and approximation of
   their spectra
229--244


Pedro Ter\'{a}n and Miguel L\'{o}pez-D\'{\i}az
Approximation of mappings with values which are upper semicontinuous
   functions
245--265


V. N. Konovalov and D. Leviatan
Kolmogorov and linear widths of weighted Sobolev-type classes on a
   finite interval, II
266--297


A. L. Levin and D. S. Lubinsky
Green equilibrium measures and representations of an external field
298--323


Note


Yuguang Xu
A note on a continuous approximate selection theorem
324--325


Erratum


Volume {\bf 112}, Number 1 (2001), in the article ``Basic analog of
   Fourier series on a $q$-linear grid'' by J. Bustoz and J. L. Cardoso,
   pages 134--157
326


Author index for Volume 113
327
------------------------------


Date: Thu, 27 Sep 2001 18:31:59 +0300
From: Eleftherios Kofidis <[e-mail]>
Subject: #23 Answer: Factoring polyphase matrices WD 10.3 #39


Dear Julian,


No, you cannot factorize any FIR polyphase matrix into lifting steps.
Indeed the two filters have to constitute a biorthogonal pair. This
means that this matrix, viewed as a matrix polynomial, must have a
determinant equal to a monomial, i.e., a constant or at most a delay
(z^{-k}). This is to ensure that its inverse (that is the synthesis
matrix) will also be FIR, i.e, polynomial. Such matrices (known as
unimodular when having constant determinant) can always be expressed as
products of elementary matrices (just as in the Gaussian elimination),
which are nothing but the lifting steps. These are in fact the so-called
ladder realizations. Based on the properties of Euclid's algorithm for
polynomial division, you can even predict what will the degrees of the
polynomials appearing in the lifting steps will be. In addition to the
papers by W. Sweldens you can have a look at the following:
M. Vetterli and C. Herley, IEEE Trans. SP, Sept. 1993,
M. Khansari and E. Dubois, IEEE Trans. SP, Aug. 1996,
and for multi-band filter banks:
E. Kofidis et al., IEEE Trans. SP, Oct. 1996,
E. Kofidis et al., EUSIPCO-96,
E. Kofidis et al., Signal Processing, July 1998.


I hope this will be of some help.


Eleftherios Kofidis
Dept of Informatics and Telecommunications
University of Athens
15784 Athens
Greece
E-mail: [e-mail]


    Date: Thu, 17 May 2001 12:21:21 +1000
    From: Julian Magarey <[e-mail]>
    Subject: #39 Question: lifting and factorising of FIR filter pairs
    
    I have a simple question about lifting.
    Is it possible to factorise the polyphase matrix from *any* FIR analysis
    
    pair into elementary lifting steps?  From reading papers I suspect not -
    
    the FIR filters may have to be biorthogonal, or some other weaker
    constraint - but I would like a definitive answer.
    
    Many thanks,
    Julian Magarey.
    
------------------------------


Date: Fri, 5 Oct 2001 17:48:23 +0200 (MEST)
From: Andreas Niedermeier <[e-mail]>
Subject: #24 Answer: Help about wavelets and images (WD 10.02 #27)


Hello Sileye-Oumar, hello waveleters


actually I just finished my PhD thesis (in german, english version
available in a few months) about a wavelet based waterline/coastline
extraction algorithm on synthetic aperture radar images of river estuaries
(exemplary using the German river Elbe). As input the original ESA SAR
amplitude images (32 bit uncompressed float data) are used, the software
runs partly in C (code from Mallat's dyadic transform at
ftp://ftp.cs.nyu.edu/pub/wave/software/) and IDL (postprocessing) on
workstations. For the moment the algorithm can be found in
A. Niedermeier, E. Romanee&szlig;en, S. Lehner: Detection of Coastlines in
SAR Images using Wavelet Methods, IEEE-GRS, Vol.38, No. 5, Sept. 2000,
pp. 2270-2281.
Concerning the image file conversions I recommend `xv' or the IDL input
readers for JPEG, GIF, etc.


Greetings,
Andreas Niedermeier,
Remote Sensing Technology Institute at German Aerospace Center,
PhD student at Dep. of Math., Munich University of Technology.
[e-mail], [e-mail]
http://www.in.tum.de/~niederma (1st hit on "Anderl Niedermeier" at Altavista)


------------------------------


Date: Tue, 25 Sep 2001 17:00:31 -0700
From: Bodo Parady <[e-mail]>
Subject: #25 Question: SPEC seeking new benchmark applications


SPEC (Standard Performance Evaluation Corporation, an
industry consortium, www.spec.org) is developing a new
CPU benchmark suite.  SPEC is seeking developers who
have key applications who can contribute source code
to SPEC for its CPU200x benchmark suite.


Applications of particular interest are those with memory
requirements of 1 GB while running and fairly long runtimes.
Real applications using wavelets would be of interest
to SPEC and its members.


SPEC is also seeking contributions to its OpenMP suites
and HPC suites if you have even larger applications.
SPEC intends to open an awards program for any
contributions.  A public announcement will be made
when details of the program are made final.


Regards


    Bodo


------------------------------


Date: Wed, 24 Oct 2001 15:30:24 +0700
From: [e-mail]
Subject: #26 Question: Echo cancellation using wavelets


Hello dears,


can somebody help me on the issue? I am doing a research on wavelet
echo cancellers. Did anybody implement echo cancellation using wavelet
transforms?  I am interested in the effectiveness measurement and
comparison with subband coding algorithms.  Some code will be
useful. Thanks in advance!


Andrey V Kuzmenko
E-mail :     [e-mail]
ICQ UIN: 54702753


------------------------------


Date: Thu, 29 Nov 2001 16:14:27 +0800
From: "zhiqiang xu" <[e-mail]>
Subject: #27 Question: Wavelets on triangulations using lifting


Hello everyone:


I was wondering if it is possible to build wavelet bases on arbitrary
triangulations using the lifting scheme? 


Thanks! 


Zhiqiang Xu
The department of applied mathematics
Dalian university of technology
Dalian,China


------------------------------


Date: Fri, 30 Nov 2001 14:04:13 +0100
From: "JPW" <[e-mail]>
Subject: #28 Question: A question about 2D FFT and wavelets


Hello,
 
I am having a problem with a 2D FFT transform. It is an function that
has an discontinuity and therefore I have the  Gibbs-Wilbraham 
phenomenon in the solution.
 
I have read that a wavelet function could improve this shortcoming.
Could you please recommend a method for a 2D function (i.e. f(x,y),
where x,y, range from +/- inf)
 
Thanks,
 
Johan
 
Dr. Johan P. Wideberg Profesor Asociado de Ingenieria Mecanica
Ingenieria e Infraestructura de los Transportes Departamento de
Mecanica y de los Materiales ESI,Universidad de Sevilla Camino de
los Descubrimientos s/n 41092 Sevilla Espana (SPAIN) Tel (of.)  :
[phone] fax : [phone] [e-mail]
 
------------------------------


Date: Wed, 12 Dec 2001 19:44:48 +0100
From: Adhemar Bultheel <[e-mail]>
Subject: #29 Question: Analysis of broadcasts of audiovisual media


For a project of the university of Antwerp we have to analyse
broadcasts of audiovisual media (radio and television).  Ideally we
should be able to semi-automatically segment the broadcast and
identify who speeks and for how long (recognizing the gender of the
speaker would be sufficient).


For the moment we are focussing on the automatic recognition of the
moment when someone starts and stops speaking
("voice-change-recognition") and the moment when a new (camera)shot
starts ("shot-change-recognition").


At first we would be happy to recognize the beginning and the end of
the intervention of a speaker. Later on we would like to also identify
the speaker and recognize the different interventions of the same
speaker.


Similarly, for the "shot-change-recognition", we only need to identify
a clear "cut" in the images, we do not want to identify people or
analyse camera-movements.


We hope that wavelets can be useful in solving these problems,
but we have no experience with this tool.
So we are looking for people who have successfully used
wavelet-transformations
(or other transformations) to analyse voice and images patterns,
but only in the limited context described above.


Concrete questions could be:
- what are the distinguishing characteristics of a voice and of a stream
of
images on which voice-change-recognition and shot-change-recognition
can be based?
- what type of wavelet transformations should be used?
- Pointers to literature where these or related problems are discussed,
will be highly appreciated.


Walter Angioletti [[e-mail]]
Nico Carpentier [[e-mail]]
University of Antwerp


------------------------------


Date: Mon, 17 Dec 2001 09:40:06 -0500
From: Chahe Nerguizian <[e-mail]>
Subject: #30 Question: Telecomm application question


Hi,


It is for a telecommunication's application. For a fixed receiver and
a mobile transmitter, and for a specific location of the transmitter
relative to the receiver, one can measure the Impulse Response (IR) or
the Power Delay Profile (PDP) of the wireless channel. I would like to
compare this measured PDP with a candidate PDP using a pattern
matching algorithm. To do so, I would like to remove the noise floor
and to extract features (de-noise and extract important information of
the PDP, such as power and delay for each multipath component) and use
the results for the pattern matching algorithm.  Is wavelet denoising
and feature extracting helpful? and if yes, what kind of information
can I extract from a wavelet? and hence what kind of wavelet should I
use (continuous, discrete, packet wavelet,...)? Any detailed
information about wavelets to be used are welcomed.  Thank you in
advance.


Chahe, PhD student


Email:[e-mail]


------------------------------


End of Wavelet Digest 10 Issue 4
************************************