| Re: Inverse CWT [message #43590] |
Wed, 20 April 2005 22:41 |
m_schellens
Messages: 31
Registered: February 2005
|
Member |
|
|
Hi Chris, Bob,
I am using WV_CWT from the wavelet package of IDL.
Basically I want to use several wavelets but for now I am using
morlets.
Then I want to delete certain frequency (scale) ranges and
resemble what is left.
The wavelet can be retrieved from the wv_fn_xxx functions.
The scales from the WV_CWT function (I need to choose /SPATIAL right?)
Anyhow, the result looks not like it should...
marc
|
|
|
|
| Re: Inverse CWT [message #43604 is a reply to message #43590] |
Wed, 20 April 2005 15:47  |
R.G. Stockwell
Messages: 363
Registered: July 1999
|
Senior Member |
|
|
"m_schellens" <mschellens@gmail.com> wrote in message
news:1113989088.320194.139730@l41g2000cwc.googlegroups.com.. .
> Does somebody know how to calculate the inverse continuous wavelet
> tansform
> in IDL?
> Thanks,
> marc
If you had used the S-Transform, you would average the local spectra over
time
to get the Fourier Spectrum (and then inverse FFT to get the time series).
Other than having that very nice quality, the ST is very similar to the CWT
with Morlet wavelets.
more info at:
http://www.cora.nwra.com/stransform/
(see the menu on the left hand side for links to code, papers, etc)
Cheers,
bob
PS sorry for not answering the question. What wavelet did you use,
and what scales do you have sampled (octave sampling?), and what
range of scales do you have? It may not be possible to get the inverse
if the full range of frequencies is not sampled.
Do you have only the CWT info and you need the time series? or
are you asking a general question on how to invert the CWT?
|
|
|
|
| Re: Inverse CWT [message #43611 is a reply to message #43604] |
Wed, 20 April 2005 12:30 |
Chris[2]
Messages: 39
Registered: August 2003
|
Member |
|
|
Hi Marc,
There are two approaches you can take. Since the wavelet transform is just a
convolution, you can use a deconvolution (or the inverse FFT if you are
doing this via FFT). However, an even simpler approach is to just sum the
real part (if complex) of the wavelet transform over all scales, and apply
an appropriate scaling factor. The scaling factor depends upon the scale,
the time sampling, and your wavelet function.
For details, I would recommend looking at Torrence and Compo, 1998: A
Practical Guide to Wavelet Analysis, available at
http://paos.colorado.edu/research/wavelets/
Hope this helps.
-Chris
Research Systems, Inc.
"m_schellens" <mschellens@gmail.com> wrote in message
news:1113989088.320194.139730@l41g2000cwc.googlegroups.com.. .
> Does somebody know how to calculate the inverse continuous wavelet
> tansform
> in IDL?
> Thanks,
> marc
>
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
