next up previous contents
Next: Adaptive filtering from Up: Noise reduction from Previous: The Wiener-like filtering

Hierarchical Wiener filtering

  In the above process, we do not use the information between the wavelet coefficients at different scales. We modify the previous algorithm by introducing a prediction of the wavelet coefficient from the upper scale. This prediction could be determined from the regression [2] between the two scales but better results are obtained when we only set to . Between the expectation coefficient and the prediction, a dispersion exists where we assume that it is a Gaussian distribution:

The relation which gives the coefficient knowing and is:

with:

and:

This follows a Gaussian distribution with a mathematical expectation:

with:

is the barycentre of the three values , , 0 with the weights , , . The particular cases are:

At each scale, by changing all the wavelet coefficients of the plane by the estimate value , we get a Hierarchical Wiener Filter. The algorithm is:

  1. Compute the wavelet transform of the data. We get .
  2. Estimate the standard deviation of the noise of the first plane from the histogram of .
  3. Set i to the index associated with the last plane: i = n
  4. Estimate the standard deviation of the noise from .
  5. where is the variance of
  6. Set to and compute the standard deviation of .
  7. i = i - 1. If i > 0 go to 4
  8. Reconstruct the picture


Rein Warmels
Mon Jan 22 15:08:15 MET 1996