** Next:** REFERENCES
** Up:** Guitton: Weighted PEF estimation
** Previous:** A 3-D land data

Estimating filters is an important step of many processing techniques
like signal/noise separation, data interpolation, missing data
restoration or regularized tomography. The stationarity of the data is
at the heart of many filter estimation techniques. In the case where
the data have strong amplitudes variations, I show that a weighting
function that leverages the fitting equations during the filter
estimation step must be used for improved results. In other words, the
residual should be weighted and not the data points. Test with 2-D and
3-D field data with stationary and non-stationary filters for two
signal-noise separation problems prove that a weighted residual PEF estimation
scheme give the best estimated noise and signal while preserving the amplitudes.

** Next:** REFERENCES
** Up:** Guitton: Weighted PEF estimation
** Previous:** A 3-D land data
Stanford Exploration Project

7/8/2003