Generalized Hampel Filters

Abstract

The standard median filter based on a symmetric moving window has only one tuning parameter: the window width. Despite this limitation, this filter has proven extremely useful and has motivated a number of extensions: weighted median filters, recursive median filters, and various cascade structures. The Hampel filter is a member of the class of decsion filters that replaces the central value in the data window with the median if it lies far enough from the median to be deemed an outlier. This filter depends on both the window width and an additional tuning parameter t, reducing to the median filter when t = 0, so it may be regarded as another median filter extension. This paper adopts this view, defining and exploring the class of generalized Hampel filters obtained by applying the median filter extensions listed above: weighted Hampel filters, recursive Hampel filters, and their cascades. An important concept introduced here is that of an implosion sequence, a signal for which generalized Hampel filter performance is independent of the threshold parameter t. These sequences are important because the added flexibility of the generalized Hampel filters offers no practical advantage for implosion sequences. Partial characterization results are presented for these sequences, as are useful relationships between root sequences for generalized Hampel filters and their median-based counterparts. To illustrate the performance of this filter class, two examples are considered: one is simulation-based, providing a basis for quantitative evaluation of signal recovery performance as a function of t, while the other is a sequence of monthly Italian industrial production index values that exhibits glaring outliers.

Pupil size as a measure of within‐task learning

“…We took a minimalist approach to cleaning the pupil data. Data that the Gazepoint GP3 system identified as poor were replaced using linear interpolation, and within‐subject outliers were corrected using a Hampel filter (see Pearson, Neubo, Astola, & Gabbouj, ). Linear interpolation was done using the zoo package (Zeileis, Grothendieck, Ryan, & Andrews, ) in R (R Development Core Team, ).…”

Solar Wind Turbulence Studies Using MMS Fast Plasma Investigation Data

“…Since we employ a low-pass filter to remove the part of the spectrum where the signal is dominated by the noise, this feature does not impact the present study. Implementation of a spectral Hampel filter and time series reconstruction: To eliminate these spikes, we adapt a decision-based filter known as Hampel filter, a well-known technique in the signal-processing community for removing outliers from a signal (Davies & Gather 1993;Liu et al 2004;Pearson et al 2016). We apply the filter in frequency spectra domain to remove the spikes, while maintaining the original phase information.…”

Predicting Remaining Useful Life with Similarity-Based Priors

“…We established experimentally that resampling the data to the minute barely loses any information from the signal, while still substantially reducing the size of the dataset. Third, to avoid the amplification of clear outliers, they are removed with a Hampel filter [14]. Fourth, we focus on the 'exponential deterioration stage' of the filter's life cycle [5], because-according to the company-the start of that stage is early enough to be able to act on time, and because it provides us with a dataset that is suitable for similarity-based RUL prediction techniques.…”