"level" parameter
171 views
Skip to first unread message
davrandom
Dec 7, 2012, 10:31:50 AM12/7/12
to pywav...@googlegroups.com
Hi,
the question is fairly simple, but since I started yesterday to program in python, I don't have a simple answer ;)
The "level" parameter in wavedec method controls the number of decomposition steps to perform. The fact is that the wavelet transform starts from the finer to the coarser resolution.
This is perfectly fine and correct when you want to encode/decode a signal.
But this is not what I want to achieve.
I am comparing the fourier series description to the wavelet description of a delta function. Easy (from the analytically point of view).
Now I want to plot this thing and here comes pywt. The result I want to achieve is more or less the one reported in the plot on figure 8 and 9 in "An Introduction to Wavelets", Lee A. Barford, R. Shane Fazzio, David R. Smith HPL-92-124
So I want to control the scale UP to with the transform is calculated, and not DOWN. (A temporary -and maybe not really clever- solution is to pass a signal with different length... from 2 samples ["low res"] up to the resolution you want)
I don't know if this is enough clear, but Thanks for your attention
d
the question is fairly simple, but since I started yesterday to program in python, I don't have a simple answer ;)
The "level" parameter in wavedec method controls the number of decomposition steps to perform. The fact is that the wavelet transform starts from the finer to the coarser resolution.
This is perfectly fine and correct when you want to encode/decode a signal.
But this is not what I want to achieve.
I am comparing the fourier series description to the wavelet description of a delta function. Easy (from the analytically point of view).
Now I want to plot this thing and here comes pywt. The result I want to achieve is more or less the one reported in the plot on figure 8 and 9 in "An Introduction to Wavelets", Lee A. Barford, R. Shane Fazzio, David R. Smith HPL-92-124
So I want to control the scale UP to with the transform is calculated, and not DOWN. (A temporary -and maybe not really clever- solution is to pass a signal with different length... from 2 samples ["low res"] up to the resolution you want)
I don't know if this is enough clear, but Thanks for your attention
d
davrandom
Dec 13, 2012, 9:47:41 AM12/13/12
to pywav...@googlegroups.com
Hi,
I managed to do this... I don't know if someone is interested... ;) Anyway the result is not astonishing, I get what I was expecting (finally).
The 'idea' in previous post was not working as expected because the representation of the wavelet resulted sparse...
So the idea is to do the full decomposition, then put to zero the higher order coefficients (i.e. 1, 2, 3, ... level coefficients) and reconstruct the signal. This way works: i get a nice wavelet well defined.
Thanks
d
Attached the comparison between the description of a delta with Fourier series and wavelet series.
I managed to do this... I don't know if someone is interested... ;) Anyway the result is not astonishing, I get what I was expecting (finally).
The 'idea' in previous post was not working as expected because the representation of the wavelet resulted sparse...
So the idea is to do the full decomposition, then put to zero the higher order coefficients (i.e. 1, 2, 3, ... level coefficients) and reconstruct the signal. This way works: i get a nice wavelet well defined.
Thanks
d
Attached the comparison between the description of a delta with Fourier series and wavelet series.
Virgil Stokes
Mar 24, 2013, 11:20:51 AM3/24/13
to pywav...@googlegroups.com
>
>
This is a nice demo --- thanks!
You seem to working a lot with the PyWavelets package --- Have you any
suggestions on using this package on unequally spaced samples (i.e. irregular
sampling)?
Best,
--V. Stokes
Reply all
Reply to author
Forward
0 new messages