Control Points¶
A control point is a known piece of information about calibration result, known prior to calibration. It can be used to improve or measure the calibration results.
Checkpoint vs Control Point¶
A control point may be used either to improve, or to measure the quality of calibration results. In the former case, it is referred to simply as control point, while in the latter case it is referred to as checkpoint.
short name |
Engine API |
goal |
|---|---|---|
control point |
|
improve calibration quality |
checkpoint |
|
measure calibration quality |
Definitions¶
control point¶
A control point is either a MTP or a GCP.
MTP¶
MTP is an acronym for Manual Tie Point. Because historically all ties points where created manually, this acronym is still used as an equivalent of tie point.
A tie point represents a set of 2D positions on images, where each position is the projection of the same 3D point in space.
There is only one 2D position per image, and only one 3D point in space.
Unless specifically stated, this 3D-point is unknown prior to calibration processing. If this 3D-point is known prior to calibration processing, please use GCP described below.
examples of tie-points¶
A tie point can be any of, but not limited to:
manual tie point
semantic tie point
chilli tie point
GCP¶
GCP is an acronym for Ground Control Point.
A GCP is a MTP of a known 3D position (together with the uncertainty on that position). It represents known projections onto different images, of the same known physical 3D position.
There is no constraint for the GCP coordinate to be on the ground.
A GCP is a type of position constrained tie point, where the known 3D position is using a known coordinate system.
Because historically all position constrained tie points where GCPs, they are still used as synonyms.