Forums

Order of Wavelets coefficients....

Started by aredo3606gif May 5, 2012
Hello,

could anyone please explain me why the order of wavelets coefficients given
on this site: 

http://wavelets.pybytes.com/wavelet/db4/

is exactly the opposite of that found on this daub.h file that includes the
Daubechies coefficients calculated by Kazuo Hatano: 

http://www.musicdsp.org/showone.php?id=20

I can see that the coefficients are already normalized so divided by
1/sqrt(2) on both. And the signs for the low pass filter case are correct
and match between the two. 

However on pybytes the order of the coefficients is exactly the opposite
from h0...h7 on Daubechies 4 (db4), the same is true for the others listed
there, to that on daub.h as calculated by Kazuo Hatano (and as it appears
in DSP books). 
What I don't get is if pybytes wrote the coefficients order in the wrong
order or if there is a reason why they are listed in reverse order there? 

Anyone could explain it ?

Thank You
On Saturday, May 5, 2012 9:46:08 AM UTC-7, aredo3606gif wrote:
> ... > However on pybytes the order of the coefficients is exactly the opposite > from h0...h7 on Daubechies 4 (db4), the same is true for the others listed > there, to that on daub.h as calculated by Kazuo Hatano (and as it appears > in DSP books). > What I don't get is if pybytes wrote the coefficients order in the wrong > order or if there is a reason why they are listed in reverse order there? >
When you apply these coefficients in a calculation do you use a dot-product routine or a convolution routine? Dale B. Dalrymple
On 5/5/12 4:28 PM, dbd wrote:
> On Saturday, May 5, 2012 9:46:08 AM UTC-7, aredo3606gif wrote: >> ... >> However on pybytes the order of the coefficients is exactly the opposite >> from h0...h7 on Daubechies 4 (db4), the same is true for the others listed >> there, to that on daub.h as calculated by Kazuo Hatano (and as it appears >> in DSP books). >> What I don't get is if pybytes wrote the coefficients order in the wrong >> order or if there is a reason why they are listed in reverse order there? >> > > When you apply these coefficients in a calculation do you use a dot-product routine or a convolution routine? >
bingo. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."