|
|
| Re: Help with WV_CWT [message #81680 is a reply to message #81654] |
Mon, 08 October 2012 18:25  |
Phillip M. Bitzer
Messages: 7
Registered: September 2012
|
Junior Member |
|
|
Balt-
AFAIK, there is no "simple" connection between frequency and scale. It's
not quite as simple as an FFT (although you might argue to ordering of
frequencies returned by IDL isn't exactly simple :-) )
There are a couple things to help you along. First, the IDL Wavelet
toolkit helped me a lot when learning about wavelets:
vis.lbl.gov/NERSC/Software/idl/help/docs6.0/wavelet.pdf
I also suggest you take a peek at the site:
http://paos.colorado.edu/research/wavelets/
Chis Torrence wrote most (all?) of the wavelet routines for IDL. The
article "A Practical Guide to Wavelet Analysis" he is lead author on is
quite useful.
For the immediate problem of how scale relates to frequency, check out
this from the aforementioned web site:
The period (or inverse frequency) is the approximate Fourier period that
corresponds to the oscillations within the wavelet. For all wavelets,
there is a one-to-one relationship between the scale and period. The
relationship can be derived by finding the wavelet transform of a pure
cosine wave with a known Fourier period, and then computing the scale at
which the wavelet power spectrum reaches its maximum.
http://paos.colorado.edu/research/wavelets/faq.html#scale
I recall that I mucked around a lot "under the hood" with the wavelet
routines to really understand them. I also use wv_applet quite a bit to
check my own code.
Hope this helps!
|
|
|
|
| Re: Help with WV_CWT [message #81681 is a reply to message #81680] |
Mon, 08 October 2012 15:53 |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On 10/8/12 4:38 PM, Balthasar Indermuehle wrote:
> Further examination of this seems to suggest that the Y scale is in
> fact contained in the returned variable "scales", however, this is an
> assumption as the frequency components it finds in the test signal
> are now closer to where they should be but still not at the exact
> spot. I find no documentation that would evidence that this is what
> it is... wavelets with IDL, the great unknown...?
Maybe you could use the "iTools help system":
$IDL_ROOT/lib/wavelet/source/wv_cwt.pro
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL: A Guide to IDL Programming (http://modernidl.idldev.com)
Research Mathematician
Tech-X Corporation
|
|
|
|
| Re: Help with WV_CWT [message #81682 is a reply to message #81681] |
Mon, 08 October 2012 15:38 |
Balthasar Indermuehle
Messages: 22
Registered: August 2012
|
Junior Member |
|
|
|
Further examination of this seems to suggest that the Y scale is in fact contained in the returned variable "scales", however, this is an assumption as the frequency components it finds in the test signal are now closer to where they should be but still not at the exact spot. I find no documentation that would evidence that this is what it is... wavelets with IDL, the great unknown...?
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
