Home Preprocessing Smoothing of Spectra
See also: Derivatives, Spatial Filtering of Images, Maximum Noise Fraction Transform, Resampling Spectra, SmoothSpectrum, Fourier Analysis of Spectra, SmoothAllSpectra, FFT Based Filtering of Spectra 

Filtering of Spectra

Command: Preprocessing > Filter > Spectral > Direct Filter

The command Preprocessing > Filter > Spectral > Direct Filter provides various algorithms for smoothing noisy spectra(1):

  • Moving average
  • Weighted average
  • Polynomial smoothing (Savitzky-Golay)
  • Moving median
  • Penalized spline
  • Weighted sinc low pass filter
  • Weighted sinc high pass filter
  • Weighted sinc band pass filter
  • Weighted sinc band stop filter
  • Chebychev low pass filter

The number of data points used for estimating the smoothed data (the smoothing window width Window Size) can be adjusted between 5 and 25 data points. Please note that the first and the last Window Size/2 data cannot be smoothed and are replaced by the nearest available smoothed value.

The following graphic shows an example of the smoothing effect on a THz spectrum using a polynomial (Savitzky-Golay) filter:

When calling the command Preprocessing > Smoothing > Spectral > Direct Filter the smoothing dialog is opened which allows to set up the parameters of the smoothing process and to have a preview of the results. The actual smoothing of all available spectra is performed only after clicking the "Apply" button .


(1) These smoothing algorithms should not be applied to mass spectra, since the implicit assumption when applying smoothing algorithms is a strong correlation of neighboring data points, which is most often not true for mass spectra. Smoothing mass spectra will destroy part of the information contained in mass spectra.