Olympic filter is the name of a digital filter for removing spikes in the data. In the "Image Processing Handbook" 7th Ed., pg 192, this name is rationalized as follows:

A modification to the simple averaging of neighborhood values that attempts to achieve some of the advantages of the median filter is the so-called Olympic filter. The Olympic filter (the name comes from the system of scoring used in some events in the Olympic games in which the highest and lowest scores are discarded and the remainder averaged) attempts to combine the median and averaging methods. By discarding the extreme values, shot noise is rejected.

Does anyone know who proposed this filter or where was its first literature occurrence? In Google Scholar the term suddenly appears in 1995, without any reference to an earlier text. Ayres, Thomas R., et al. "The RIASS coronathon: Joint X-ray and ultraviolet observations of normal FK stars." The Astrophysical Journal Supplement Series 96 (1995): 223-259.

If more than five profiles were available we applied an "Olympic" filter, discarding the high and low values in each wavelength bin.

I don't think this is the correct earliest reference. I had posted in SE Science & Math History earlier but it did not receive an answer.

  • 2
    $\begingroup$ This is a trimmed or truncated mean. I’m sure it’s been used as a filter before 1995, maybe the authors of this paper are the first ones to call it “Olympic”, maybe as a joke, or maybe it was a name running around their department but not in common use elsewhere. $\endgroup$ Oct 2, 2021 at 23:38
  • 1
    $\begingroup$ If the number of points (or "values") is large enough, you may want to kick out more than the one top and one bottom scores and average the rest. Let's say it's a sliding window of 9 values, kick out the two top and two bottom and average the remaining five. $\endgroup$ Oct 3, 2021 at 2:57


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.