get_component_relations🔗
Short description🔗
get_component_relations — Return the relations between the model components that are contained in
a training result.
Warning🔗
get_component_relations is obsolete and is only provided for
reasons of backward compatibility.
The operator will be removed with HALCON 26.11.
Signature🔗
get_component_relations( out region Relations, component_training ComponentTrainingID, integer ReferenceComponent, string Image, out rectangle2.center.y Row, out rectangle2.center.x Column, out rectangle2.angle.rad Phi, out rectangle2.hwidth Length1, out rectangle2.hheight Length2, out angle.rad AngleStart, out angle.rad AngleExtent )
Description🔗
get_component_relations returns the relations between model
components after training them with train_model_components.
With the parameter ReferenceComponent, you can select a
reference component. get_component_relations then returns
the relations between the reference component and all other components
in the model image (if Image \(=\) 'model_image' or
Image \(=\) 0) or in a training image (if
Image >= 1). In order to obtain the
relations in the ith training image, Image must be
set to i. The result of the training returned by
train_model_components must be passed in
ComponentTrainingID. ReferenceComponent describes
the index of the reference component and must be within the range of
0 and n-1, if n is the number of model components
(see train_model_components).
The relations are returned in form of regions in Relations
as well as in form of numerical values in Row,
Column, Phi, Length1, Length2,
AngleStart, and AngleExtent.
The region object tuple Relations is designed as
follows. For each component a separate region is
returned. Consequently, Relations contains n
regions, where the order of the regions within the tuple is
determined by the index of the corresponding components. The
positions of all components in the image are represented by circles
with a radius of 3 pixels. For each component other than the
reference component ReferenceComponent, additionally the
position relation and the orientation relation relative to the
reference component are represented. The position relation is
represented by a rectangle and the orientation relation is represented
by a circle sector with a radius of 30 pixels. The center of the
circle is placed at the mean relative position of the component. The
rectangle describes the movement of the reference point of the
respective component relative to the pose of the reference
component, while the circle sector describes the variation of the
relative orientation (see train_model_components). A
relative orientation of 0 corresponds to the relative orientation of
both components in the model image. If both components appear in the
same relative orientation in all images, the circle sector
consequently degenerates to a straight line.
In addition to the region object tuple Relations, the
relations are also returned in form of numerical values in
Row, Column, Phi, Length1,
Length2, AngleStart, and
AngleExtent. These parameters are tuples of length n
and contain the relations of all components relative to the
reference component, where the order of the values within the tuples
is determined by the index of the corresponding component. The
position relation is described by the parameters of the corresponding
rectangle Row, Column, Phi,
Length1, and Length2 (see
gen_rectangle2). The orientation relation is described by
the starting angle AngleStart and the angle extent
AngleExtent. For the reference component only the position
within the image is returned in Row and
Column. All other values are set to 0.
If the reference component has not been found in the current image, an array of empty regions is returned and the corresponding parameter values are set to 0.
The operator get_component_relations is particularly useful
in order to visualize the result of the training that was performed
with train_model_components. With this, it is possible to
evaluate the variations that are contained in the training
images. Sometimes it might be reasonable to restart the training
with train_model_components while using a different set of
training images.
Execution information🔗
Execution information
-
Multithreading type: reentrant (runs in parallel with non-exclusive operators).
-
Multithreading scope: global (may be called from any thread).
-
Processed without parallelization.
Parameters🔗
Relations (output_object) region(-array) → object
Region representation of the relations.
ComponentTrainingID (input_control) component_training → (handle)
Handle of the training result.
ReferenceComponent (input_control) integer → (integer)
Index of reference component.
Restriction: ReferenceComponent >= 0
Image (input_control) string → (string / integer)
Image for which the component relations are to be returned.
Default: 'model_image'
Suggested values: 'model_image', 0, 1, 2, 3, 4, 5, 6, 7, 8
Row (output_control) rectangle2.center.y(-array) → (real)
Row coordinate of the center of the rectangle representing the relation.
Column (output_control) rectangle2.center.x(-array) → (real)
Column index of the center of the rectangle representing the relation.
Phi (output_control) rectangle2.angle.rad(-array) → (real)
Orientation of the rectangle representing the relation (radians).
Assertion: -pi / 2 < Phi && Phi <= pi / 2
Length1 (output_control) rectangle2.hwidth(-array) → (real)
First radius (half length) of the rectangle representing the relation.
Assertion: Length1 >= 0.0
Length2 (output_control) rectangle2.hheight(-array) → (real)
Second radius (half width) of the rectangle representing the relation.
Assertion: Length2 >= 0.0 && Length2 <= Length1
AngleStart (output_control) angle.rad(-array) → (real)
Smallest relative orientation angle.
AngleExtent (output_control) angle.rad(-array) → (real)
Extent of the relative orientation angles.
Result🔗
If the handle of the training result is valid, the operator
get_component_relations returns the value 2 (H_MSG_TRUE).
If necessary, an exception is raised.
Combinations with other operators🔗
Module🔗
Matching