How can I use wavelet thresholding with nondecimated 2D decomposition of an MR image ?

I am applying Bilateral Filter to the approximations and I want to soft threshold the details from level 3 ..How can I implement wthcoef2 for the non decimated wavelet decomposition structure ?

data = load('mri'); % MRI dataset (magnitude image) distributed with MATLAB
image_num=15;
im = double(data.D(:,:,1,image_num)); % image_num max=27
s=5;
im_r = rician(im, s); % Add Rician noise (make Rician distributed)

% Compute ranges
min_o = round(min(im(:)));  
max_o = round(max(im(:)));
min_r = round(min(im_r(:))); 
max_r = round(max(im_r(:)));

s = 0.05*max_o; % Percentage of NOISE LEVEL
im_R=rician(im,s);
w='haar';
NBCOL = size(data.map,1);
n=3;

%Multilevel Undecimated Discrete 2D Wavelet Transform

WT  = ndwt2(im_R,n,w);    

A = cell(1,n);
D = cell(1,n);

for k = 1:n
    A{k} = indwt2(WT,'a',k);   % Approximations 
    D{k} = indwt2(WT,'d',k);   % Details
end

DH=[];
for k=1:n
    D1=D{k};
    D1=D1(:);
    DH=[DH D1];
end
delta=median(abs(DH))/0.6745;
thr=delta * sqrt(2*log(n));
NC = wthcoef2('t',WT.dec,WT.sizes,1,thr,'s');

What should I use instead of WT.dec and WT.sizes ?

Answer 1

[Wayne King]

You want to use sw2dtool

>>sw2dtool

Interactively play with your denoising and then select

File -> Generate MATLAB Code (Denoising Process)

That will generate the MATLAB code to show you how to do it from the command line.

Answer 2

[Anjoo Patel]

Image decomposition