Noise estimation and thresholding methods in Discrete Wavelet Transform

Started by sh_ee 4 years ago2 replieslatest reply 4 years ago139 views

I am a bit confused regarding the thresholding methods and noise estimation options found in Matlab regarding Discrete Wavelet Transform.

Please correct me if I am wrong.

I am understanding that noise estimation options are estimating the noise per scale (level dependent) or estimating the noise per the first level (level independent). Reading about thresholding methods, SURE or Bayes; for example, we have the option to estimate noise level dependent and level independent with both these methods. But both of these methods are considered adaptive or subband. So would this adaptability be only true if the level dependent noise estimation is chosen otherwise it will act as a fixed threshold but uses its relevant method of calculating the appropriate threshold?

Thank you

[ - ]
Reply by omersayliSeptember 16, 2020


Even if it is fixed, you apply in on sub-bands (there could be coefficients exceeding threshold or not). Coefficients are found for each level of tree, hence adaptive or sub-band processing becomes again. 

[ - ]
Reply by sh_eeSeptember 16, 2020

Thank you Omersayli for replying,

I have two question then, 

In level independent all the threshold levels are the same number (one number), while level dependent contains a different threshold level per that particular level (an array)?

So all threshold methods are considered adaptive, even the universal threshold?