Re: Simplifies wavelet denoising
From: illwhacker (illywhacker_at_free.fr)
Date: 02/26/05
- Previous message: Mattd: "Re: Object position in camera view"
- In reply to: pslant: "Simplifies wavelet denoising"
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 26 Feb 2005 14:35:26 +0100
If the noise is white Gaussian, it occurs equally at all frequencies.
Visually, though, the lower frequencies might be less noticeable/bothersome
than the high frequencies, so denoising only the high frequencies might be
all that you need. If you have implemented the first stage of the wavelet
decomposition, however, the effort to make it recursive is very small, so
you might as well do it and compare the results.
illywhacker;
"pslant" <pslant@yahoo.com> wrote in message
news:40b9bb91.0502250949.7afbd77@posting.google.com...
> I'm trying to understand wavelets by implementing wavelet denoising.
> I havent yet implemented the recursive chain, only "subdivision" of
> the original images into 4 parts with the upper left being the low
> pass,downsampled, version of the original.
> My question is this: is it really necessary to go any further for
> denoising?
> the other 3 quadrants contains all the High pass stuff where you most
> likely will find any noise?
> Wouldnt it be sufficient to threshold these 3 parts and than
> reconstruct the image?
> Is this a correct supposition?
- Previous message: Mattd: "Re: Object position in camera view"
- In reply to: pslant: "Simplifies wavelet denoising"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
