www.pudn.com > multiwavelet.rar > thresh_vec2D.m

function transfth=thresh_vec2D(transf,maxlevel,thresh,covname,threshtype) 
%transfth=thresh_vec2D(transf,maxlevel,thresh,covname,threshtype) 
% 
%  This function performs vector thresholding with decorrelation of  
%  2-dimensional multiwavelet transform. Scaling coefficients stay untouched.  
%  For description of the algorithm see [DS], [SW]. 
%  
%  Input:   
%    threshtype  string of characters, either 'hard' or 'soft'; 
%                  threshtype='hard' invokes hard thresholding, 
%                  threshtype='soft' invokes soft thresholding; 
%                for details see [DS] 
%    covname     string of characters, name of covariance matrix used for  
%                decorrelation; for possible names and short descriptions  
%                see  coef_dcov2D.m 
%    thresh      real, the value of the threshold 
%    maxlevel    integer, the number of levels of wavelet decomposition  
%                in transf, maxlevel < 6 
%    transf      n by n real array, wavelet transform of a signal; 
%                n must be of the form integer*2^maxlevel; 
%                for structure of transf see [SW] 
% 
%  Output:  
%    transfth    n by n real array, thresholded multiwavelet coefficients 
% 
%  Example of Usage: 
%    transfth=thresh_vec2D(transf,3,0.75,'ghmap','hard') 
 
% Author: Vasily Strela 
% COPYRIGHT 1997,98 by Vasily Strela 
 
n0=length(transf(1,:)); 
[HH,HL,LH]=coef_dcov2D(covname); 
r=length(HH(:,1)); 
nf=round(sqrt(r)); 
 
for i=1:maxlevel, 
  hh=inv(HH(:,(i-1)*r+1:i*r)); 
  hl=inv(HL(:,(i-1)*r+1:i*r)); 
  lh=inv(LH(:,(i-1)*r+1:i*r)); 
  nb=n0-(2^i-1)*n0/2^i; 
  ste=round(n0/(nf*2^i)); 
 
  for j=nb+1:nb+ste, 
    for k=nb+1:nb+ste, 
      tj=zeros(r,1); 
      for m1=1:nf 
        for m2=1:nf 
          tj((m1-1)*nf+m2)=transf(j+(m1-1)*ste,k+(m2-1)*ste); 
        end 
      end 
      val=sqrt(tj'*hh*tj); 
      if val