Outlier Rejection

Outlier rejection allows filtering of data points based on feature type geometry. Data for a circle feature can be filtered relative to the fit of a circle solution. Likewise, data for a line feature can be filtered relative to the fit of a line to the data.

Outlier rejection is available for the following features:

Point	Line	Circle	Arc
Ellipse	Plane	Gap	Rectangle
Slot	Sphere	Cylinder	Cone

The outlier rejection can be accessed from the rightclick menu in the Features List, Schematic View, Log View, Points View, and through the Feature menu.

Access and behavior are similar to the Tolerances and Exports windows. Outlier Rejection can be set for a selected feature or set for multiple selected features by feature type.

Data points can be filtered by an absolute Distance, a Standard Deviation, and a Percentage of Points.

Distance: Enter the permitted minimum and maximum deviation from the fit. Points exceeding the minimum or maximum values will be rejected. Note that the units are the same as the units set up in InSpec Software.

Standard Deviation: Enter the standard deviation that will limit the data. Any points that fall outside the set standard deviation will be rejected.

Users should look at the graph before setting a threshold, keep in mind that not all measured features will follow an approximately normal distribution.

Use 2nd Threshold filtering: When the 2nd Threshold filtering is used, the points outside of the Standard Deviation threshold are removed as usual. Additionally, any points that are connected to the rejected points will be removed down to the 2nd threshold value. When filtering detects an increase in deviation, the filtering will terminate at the local minimum. A local minimum can indicate that an independent outlier is present. Multiple iterations can remove multiple independent outliers.

Percentage of Points: Users can move the sliders right and left to adjust the percentage of points that should be considered as part of the measurement. i.e.: sliding the bar to +95 %, means that:

5 % of the total data will be removed.
The points removed are the ones with the largest positive deviation from the fit.

Max. Iterations: After each iteration of outlier rejection, the Feature geometry will be refit to the remaining points and a new Standard Deviation will be calculated.

When Distance or Standard Deviation is selected, the operator can determine the maximum amount of iterations.

The rejected points will be colored gray instead of black in the Points View window and the line connecting the point and the feature solution will not be drawn

In the Playback Options Failure tab, there is now a third option to “Pause program if a Feature fails because of Outlier Rejection”.

This line was also added to the New Program Defaults- Run Failures tab.