wavelet threshold select ddencmp
2 views (last 30 days)
Show older comments
1.hello i want to denoise image using wavelets i think about use wavelet toolbox i want to use ddencmp but as i understand for denoise it use soft threshold always how to set hard threshold
2. i want to take only several sub images from wavelet decomposition to restore original image how to do it ? (after threshold i determine)
0 Comments
Answers (2)
Wayne King
on 29 Jul 2012
Hi ddencmp only returns thresholds so there is nothing "soft" or "hard" about it. When you actually denoise the images using wdencmp, then you want to pass the option for sorh as 's'.
You can try this out easily through the wavemenu GUI, by selecting Wavelet 2-D, importing your image, and then selecting De-Noise.
If you only want to reconstruct an approximation from certain details, you can use waverec2.m
0 Comments
michael scheinfeild
on 27 Aug 2012
1 Comment
Apoio
on 10 Jul 2014
I have the same question. I tried to follow the wavemenucode (sw2dtool) and I cant find where thresholds are determined, (one gets lost on the code...).
I also tried:
wname = 'haar';
level = 5;
OI=imread('test.png');
[C,S] = wavedec2(OI,level,wname);
THR = wthrmngr('dw2ddenoLVL','sqtwolog',C,S,'mln')
(the parameters ('dw2ddenoLVL','sqtwolog','mln') are as suggested by the code for denoising generated by wavemenu)
and the resulting threshold values are very different from the ones on the wavemenu GUI (and on automatically generated code).
Since the denoised image looks much better using the thresholds given by wavemenu GUI than the ones given by wthrmngr, I'd like to find out how they are computed...
TIA
See Also
Categories
Find more on Denoising and Compression in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!