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 6, Issue 5


Preprint: Wavelets optimized according to perceptual criteria
 
images/spacer.gifimages/spacer.gif Reply into Digest
Previous :: Next  
Author Message
"Prof. Eduardo" (eduardo@coe.ufrj.br)
Guest





PostPosted: Tue Apr 22, 1997 2:17 pm    
Subject: Preprint: Wavelets optimized according to perceptual criteria
Reply with quote

#3 Preprint: Wavelets optimized according to perceptual criteria

Dear Waveletters,

We are conducting a research on wavelet optimization for image
processing, using a perceptual criteria that combines coding gain
performance with ringing effect minimization. Wavelets are
biorthogonal and filters that generate them are linear-phase. In
the course of our work we came across a filter bank whose filters
are of length 6 each. They prove to be both small in length and
have very good performance for image coding. If some of the readers
want to experiment with them, the coefficients of the analysis
lowpass filter are:

h0 = [-0.023447261264697;-0.052757123924977;0.78331116637622;...
0.78331116637622;-0.052757123924977;-0.023447261264697]

And the coefficients of the analysis highpass filter are:

h1 = [-0.050152151183346;-0.11284402153473;0.64441491083516;...
-0.64441491083516;0.11284402153473;0.050152151183346]

We have compared it to filter 4 (analysis lowpass of length 5 and
highpass of length 3) in Villasenor et. al., ("Wavelet filter
evaluation for image compression", IEEE Trans. on Image Processing,
Aug 95, vol.4, no. 8, pp.1053-1060) and to filter 3 (analysis
lowpass of length 6 and highpass of length 10) in the same
reference. In the first case our filter have a superior performance
and in the second a comparable, except for the ringing, that was
very small with our filter. These comparisons were made using EZW
algorithm at 0.15 bpp on the 512X512 LENA image. Note that since
the filters are biorthogonal, the bands should be scaled according
to the method described in J. H. Woods and T. Naveen, "A filter
based bit allocation scheme for subband compression of HDTV", IEEE
Trans. on Image Processing, vol.1, no. 3, Jul 1992.

We would appreciate if those involved in image processing would use
this wavelet and return some comments to us.

More details can be found on the following references:

da Silva, E. A. B. and Ghanbari, M., "On the performance of linear
phase wavelet transforms in low bit-rate image coding", IEEE Trans.
on Image Processing, vol. 5, no. 5, pp. 689-704, May 1996, and

Rodrigues, M. A. M., da Silva, E. A. B. and Diniz, P. S. R.,
"Design of wavelets for image compression satisfying perceptual
criteria", Electronics Letters, vol. 33, no. 1, pp. 40-41, Jan 1997.

Yours
Marco Antonio M. Rodrigues mamr@fund.cepel.br
Eduardo Antonio Barros Silva eduardo@coe.ufrj.br
Paulo Sergio R. Diniz diniz@coe.ufrj.br

http://www.coe.ufrj.br/~sergioln/profs/eduardo/
http://www.coe.ufrj.br/~sergioln/profs/diniz/
All times are GMT + 1 Hour
Page 1 of 1

 
Jump to: 
 


disclaimer - webmaster@wavelet.org
Powered by phpBB

This page was created in 0.025261 seconds : 18 queries executed : GZIP compression disabled