Image Denoising and Wavelet Thresholding

Hello,

I am currently doing a research on the above subject. I am not sure
about the level of decomposition during wavelet transform. What is the
reasonable number of levels to decompose for image denoising?

Thanks.

On 16 Srp, 13:38, Bumbala <bumb...@gmail.com> wrote:
> Hello,
>
> I am currently doing a research on the above subject. I am not sure
> about the level of decomposition during wavelet transform. What is the
> reasonable number of levels to decompose forimagedenoising?
>
> Thanks.

Hi there,
 it depends on wavelet you use. I used BayesShrink method with
daubechies7 wavelets and the level 5 was enough. Btw wavelet denoising
is not the state-of-the-art. Better is for example NonLocalMeans. My
implementation is at www.adptools.com

Filip

>On 16 Srp, 13:38, Bumbala <bumb...@gmail.com> wrote:
>> Hello,
>>
>> I am currently doing a research on the above subject. I am not sure
>> about the level of decomposition during wavelet transform. What is the
>> reasonable number of levels to decompose forimagedenoising?
>>
>> Thanks.
>
>Hi there,
> it depends on wavelet you use. I used BayesShrink method with
>daubechies7 wavelets and the level 5 was enough. Btw wavelet denoising
>is not the state-of-the-art. Better is for example NonLocalMeans. My
>implementation is at www.adptools.com
>
>Filip
>
Hi
Please do trail and error method until you get minimum Mean Square
Error.Once you cant get more less than previous MSE then stop at that
level. Different will work differently, for different wavelets it is
different.

Thanks
Naresh

> Hi
> Please do trail and error method until you get minimum Mean Square
> Error.Once you cant get more less than previous MSE then stop at that
> level. Different will work differently, for different wavelets it is
> different.
>
> Thanks
> Naresh

Hi, do you want to minimize MSE to what? MSE is zero to noised image
and you dont know denoised image ... Well, I dont really understand
what you mean.

F.