The Wavelet Digest Homepage
Return to the homepage
Search the complete Wavelet Digest database
Help about the Wavelet Digest mailing list
About the Wavelet Digest
The Digest The Community
 Latest Issue  Back Issues  Events  Gallery
The Wavelet Digest
   -> Volume 2, Issue 5


Question: Implementing C. Chui's semi-orthogonal wavelets.
 
images/spacer.gifimages/spacer.gif Reply into Digest
Previous :: Next  
Author Message
Patrick Marais
Guest





PostPosted: Fri Nov 29, 2002 2:28 pm    
Subject: Question: Implementing C. Chui's semi-orthogonal wavelets.
Reply with quote

Question: Implementing C. Chui's semi-orthogonal wavelets.

Hi,

I am currently implementing C. Chui's semi-orthogonal wavelet
decomposition (of images) as part of my master's thesis. I was able to
implement the quadratic scheme without any real problems. However, when
I included code to perform the cubic decomposition, the reconstructed
images were overlayed with a regular grid of lines in some circumstances
as outlined below.

When the number of terms in the detail decomposition sequence,
conventionally named {b}, is increased to about 25, these lines are no
longer evident. At first, I believed my boundary conditions to be at
fault; subsequently, I tried substituting an alternative sequence for
the detail sequence - just the original sequence {b} negated - without
altering my boundary conditions. Obviously, the image was distorted (as
I had expected), however, these lines are no longer visible.

It is possible that my boundary conditions may indeed be in error.
Could some kind soul out there inform me of the methods generally used
to deal with such boundary problems? I have used a periodic
symmetrization to extend the image (fairly standard as far as I know).
Furthermore, I generally only consider image (basis) coefficients
supported on a lattice marginally larger than the original image - this
certainly proved sufficient in the quadratic case. (I have, however,
tried extending this support dramatically for the cubic case, to no
avail). Is there anything else I might consider?

Any help would be greatly appreciated!
Thanks in advance,
Patrick C. Marais

e-mail: patrick@cs.uct.ac.za
All times are GMT + 1 Hour
Page 1 of 1

 
Jump to: 
 


disclaimer - webmaster@wavelet.org
Powered by phpBB

This page was created in 0.025325 seconds : 18 queries executed : GZIP compression disabled