load_wavelet.m

小波压缩编码平台软件包

function wvf=load_wavelet(wavelet,normw)
%wvf = load_wavelet(wavelet,type,normw);
%Version: 3.10, Date: 2006/01/31, author: Nikola Sprljan
%Loads properties of the specified wavelet filter
%
%Input: 
% wavelet - wavelet identification string
% normw - specifies DC and Nyquist normalisation of the wavelet, if it has
%         value 'E' the normalisation that produces a unit energy gain of 
%         synthesis filters will be performed
%
%Output: 
% wvf - structure containing properties of a wavelet:
%  'id' - identifier, i.e. a string descriptor of a wavelet 
%  'wvf_type' - symmetry type of a wavelet, can be 'symmetric_odd',
%   'symmetric_even', 'non_symmetric'. 'odd' specifies an odd number of
%   coefficients in a filter, and 'even' an even number
%  'lift_coeff' - lifting coefficients. If wavelet is not symmetric, there
%   can be two coefficients per lifting step.
%  'lift_cnct' - which of the to neighbouring pixels are used in the 
%   lifting step.
%  'lift_norm' - normalisation factors used in lifting to achieve the DC
%   and Nyquist gains of sqrt(2).
%  'filt_AL','filt_AL_delay' - analysis low-pass filter, with delays
%  'filt_AH','filt_AH_delay' - analysis high-pass filter, with delays
%  'filt_SL','filt_SL_delay' - synthesis low-pass filter, with delays
%  'filt_SH','filt_SH_delay' - synthesis high-pass filter, with delays
%
%Note:
% Decomposition <=> Analysis
% Reconstruction <=> Synthesis
% It is assumed that DC and Nyquist gain factor of the wavelet being 
% retrieved from a file (or in other two ways) equals sqrt(2). Lifting
% and filterbank coefficients must correspond in that regard.
% Wavelets specifications are stored in the directory .\Wavelets, each in 
% a human-readable .wvf file. The syntax of a .wvf file is as follows:
%   
% %multiple-line comment