Digital Signal Processing Reference
In-Depth Information
% Sort, threshold and reconstruct.
swc = sort(abs(wc(:)),1,'descend');
thd = swc(floor(N*N/10));
ll = wc(1:coarsest,1:coarsest);
wct = wc .* (abs(wc) > thd);
wct(1:coarsest,1:coarsest) = ll;
% Display retained DWT coeffs.
subplot(3,4,10);imagesc(wct~=0);axis image;axis off;
set(gca,'FontSize',14);
title('(d)');
% Reconstruct and display.
imdwtr = IWT2_PO(wct,coarsest,qmf);
subplot(3,4,11);imagesc(imdwtr);axis image;axis off;
set(gca,'FontSize',14);
title('(e)');
2.10 SUMMARY
We have used the CWT as the basic schema from which various discrete wavelet
transforms can be related. We have seen how wavelet filtering is carried out. A
number of examples demonstrated the usefulness of filtering data in this way. In
particular, the JPEG 2000 still image compression format uses a discrete wavelet
approach. MATLAB code is used to illustrate the algorithms that are discussed.
