The Filter CHOP smooths or sharpens the input channels. It filters by combining each sample and a range of its neighbor samples to set the new value of that sample. Each filter type uses its own weighting factors for the neighboring samples. The Filter Width determines the number of neighbors to use.
For a similar, more-abrupt effect, see the Lag CHOP.
Type - There are seven types of filters:
- Gaussian - This filter has a Gaussian (normal or "bell" curve) shape that smooths the channel. It acts as a low pass filter. The wider the filter, the lower the cutoff frequency, resulting in smoother data.
- Left Half Gaussian - This produces a lag on the channel. If the input channels represent values over time, this filter is seen as only using samples back in time from the current sample. For time-data, this is more realistic as you can't look ahead in time. (Maybe some day.) It has a half-bell shape.
- Box - This filter is box-shaped, meaning that each neighbor sample it uses has the same weighting factor. It can produce unwanted steps in the output channel because the effect of the samples at the extremes of the filter don't fade out as the window slides over the samples. It low-pass filters data, similar to the Gaussian filter.
- NOTE: When using Gaussian or Box filtering, the channel is delayed by half the filter size (i.e. a Filter Size of 30 samples will delay the output by 15 samples). To eliminate this delay, use either a Left Half Gaussian or a Left Half Box filter. Applying a Sharpen or Edge Detect filter always delays the output by half the filter size. Applying a Despike filter will delay the output by the full filter size.
- Left Half Box - This filter produces a lag on the data, uses only samples back in time, and otherwise acts like a box filter.
- Edge Detect - This filter detects "edges", sharp changes in the input channels. It acts as a high pass filter. As the filter width is increased, more low frequencies are added.
- Sharpen - This filter sharpens all high frequencies. It is the sum of the edge detect result and the original data.
- De-spike - This filter removes "spikes" (samples more than `Spike Tolerance' above or below the expected sample value). The filter width allows you to eliminate spikes that are several samples long. Wide filters will remove wide spikes (spikes of several samples) and small filters will only remove narrow spikes (one or two samples in length).
- Ramp Preserve - This filter attempts to output an increasing ramp that increases at the 'Ramp Rate' parameter rate. The input channel to the Filter CHOP should be increasing at this rate, with possibly some errors/noise in it. The Ramp Preserve will ignore the input channel unless the difference between the input value and the current ramp value is larger than 'Ramp Tolerance' parameter. When the difference becomes greater than the tolerance, the ramp will reset to start at the current input value.
Effect - The extent to which the filter affects the channel (0 - not at all, 1 - maximum effect).
Filter Width - The amount of surrounding samples used in the calculation of the current sample. It is expressed in the Units.
Spike Tolerance - For the De-spike filter type, this is the amount that a sample can differ from its neighbours without being considered a spike.
Ramp Tolerance - When using a Ramp Preserve filter, if the input value deviates from the current output ramp value by this much, then the ramp will reset to the new input value. Otherwise the Ramp Preserve will continue climping at the specified 'Ramp Rate'.
Ramp Rate - When using a Ramp Preserve filter, this is the rate that the CHOP's output channel will increase. Only if the input channel value deviates from the desired output value by 'Ramp Tolerance' amount will the CHOP instead output the input channel value.
Number of Passes - The number of times the filter is applied to the channel.
Filter Per Sample
/lagsamples - Applies the filter to each sample of the channel instead of across the whole channel. Useful for working with multi-sample channels.
/reset - Resets the filter.
Standard Options and Local Variables