Output Point Spacing: Where possible, this setting tries to generate the output feature with points at this spacing. However, its not guaranteed.
Cluster Radius: Where data can have gaps in it, for instance hanging cables, this value helps reconnect data so the search routine can continue. This distance specifies the max distance between any two clusters that could be considered the same cluster. A value of 0 will turn off the clustering altogether, so that should only be done with dense data sets.
Dz Tol: This value controls how far sample data can be vertically offset from the current direction vector. Imagine a line running along the top of a Kerb. Data inside this tolerance vertically will be allowed though to the next stage of the routine.
Hz Tol: This value controls how far sample data can be horizontally offset from the current direction vector. Imagine a line running along the top of a Kerb. Data inside this tolerance horizontally will be allowed though to the next stage of the routine. Larger values are usually necessary if the Kerb changes direction rapidly, to make sure enough of the Kerb is seen to establish a rough idea of where it is heading.
Search Distance: Defines how long a sample is analysed each time. For long linear highways that do not change direction sharply, larger values can be used, say 4m. For tighter twisty data, or data that can change direction quickly, smaller values will be needed, say 0.5m.
Redirect Length: Defines the distance each redirection increases its search distance by.
Max Redirects: This value sets the maximum number of tries that should be attempted if no new data is found to continue the extraction. With each redirect the routine will look further and further which helps to jump forward when there may be a shadow or gap in the date.
Grid Spacing: This value defines a cubic volume that scan data will be de-resolved down to. Smaller values increase accuracy, but as more data is used, more time is needed to process the results.
Douglas Peucker Tolerance: To help simplify the output data this value controls how aggressively resulting line work is smoothed. Larger values result in more data being removed.
Min Coeff: The Correlation Coefficient defines how strict the data being examined during each step of the extraction must match a trend. If the source data is very noisy or patchy, then values as low as 0.8 can allow extraction to occur. If the data is very well defined however, say cables that can only hang in a straight line. Then values above 0.95 can be considered.
Min SD: In essence this controls how far data points can be from the trend line in meters, before they are disregarded. So cables would need a small value, say 0.05. However, road markings may be up to 0.15 so that the entire thickness of the marking is kept.
Get Mean Height Over Length: This value is used to try and mean out noisy changes in height. Small values say <=0.1m indicate the data is dense and reliable. Larger values, e.g. 0.5m would be used for noisy photogrammerty data perhaps.
Extract Catenary: Applies to cable extraction only. This mode which is available per seed tries to fit vertical curves to the extracted data. Those curves are then used to better home in on where the cables are supported, and to remove points that are considered not on the best fit vertical curve.
Use Highs: Applies to cable extraction only. This mode which is available per seed will prefer higher data. So where two cables meet vertically to go under an obstruction for instance, or where one supporting wire meets another. The routine will follow the correct route once the cables part company.