Geopsy: Filter
Filter is a Basic signal processing tool in the Waveform menu and a tab in some advanced signal processing tools. In the first case, it applies a filter to the signals in the active signal viewer. In the second case, it applies a filter to the active signals for various purposes as detailed for each particular tool.
In Waveform menu, by clicking on Filter, a pop-up window appears to customize the filter. In that case Filter creates new temporary signals in the active viewer.
Contents
Filter type
There are four types of filter:
- Low pass filters the frequencies above a given value.
- High pass filters the frequencies below a given value.
- Band pass filters the frequencies out of a given frequency range.
- Band reject filters the frequencies inside a given frequency range.
The cut-off frequency(ies) of the filter should be provided in the neighboring text box(es). All values are in Hertz.
Filter method
Two filter methods are implemented in Geopsy:
- Butterworth filter
- Convolution by a window function
Butterworth filter
Butterworth filter is linear continuous-time filter with a maximally flat frequency response. By default, the filter is applied in both directions that makes it non-outphasing but not causal (filter output depends also on inputs in time). The Causal checkbox ensures the filter will be causal (filter output depends only on past and present inputs, so that no artificial effect can be created before the arrival of a propagating wave, for example) by an application in only one direction. However it does not conserve the phase of the signal (outphasing filter). This option should therefore not be used for processing based on phase information, e.g. array processing.
The Order of the filter defines the number of data points used for the computation of a single filtered point. Low orders less affect the signal in time but make the filter less efficient in frequency. High orders make the filter more efficient in frequency but affect more the signal in time. The default value is 2.
Convolution by a window function
Fifteen convolutions are available.
Blue: Time domain
Orange: Frequency domain
More informations from Wikipedia on window functions.
Parameters depend upon the method (to be deepen). The relative width of the window on which the filter is applied has always to be entered.
Other parameters are required for three of the filters:
- Tukey: width (alpha/2)
- Gaussian: width (sigma)
- Blackman: alpha
