function f = comp_iufwt(c,g,a,J,scaling)
%COMP_IUFWT Compute Inverse Undecimated DWT
% Usage: f = comp_iufwt(c,g,J,a);
%
% Input parameters:
% c : L*M*W array of coefficients, M=J*(filtNo-1)+1.
% g : Synthesis wavelet filters-Cell-array of length *filtNo*.
% J : Number of filterbank iterations.
% a : Upsampling factors - array of length *filtNo*.
%
% Output parameters:
% f : Reconstructed data - L*W array.
%
%
% see comp_fwt for explanantion
assert(a(1)==a(2),'First two elements of a are not equal. Such wavelet filterbank is not suported.');
% For holding the impulse responses.
filtNo = length(g);
gOffset = cellfun(@(gEl) gEl.offset,g(:));
% Optionally scale the filters
g = comp_filterbankscale(g(:),a(:),scaling);
%Change format to a matrix
gMat = cell2mat(cellfun(@(gEl) gEl.h(:),g(:)','UniformOutput',0));
% Read top-level appr. coefficients.
ca = squeeze(c(:,1,:));
cRunPtr = 2;
for jj=1:J
% Current iteration filter upsampling factor.
filtUps = a(1)^(J-jj);
% Zero index position of the upsampled filetrs.
offset = filtUps.*gOffset ;%+ filtUps;
% Run the filterbank
ca=comp_iatrousfilterbank_td([reshape(ca,size(ca,1),1,size(ca,2)),...
c(:,cRunPtr:cRunPtr+filtNo-2,:)],gMat,filtUps,offset);
% Bookkeeping
cRunPtr = cRunPtr + filtNo -1;
end
% Copy to the output.
f = ca;