set_shape_model_metric — Set the metric of a shape model that was created from XLD contours.
set_shape_model_metric changes the value of the parameter
Metric that was selected during the creation of the shape model
ModelID from XLD contours. For this, the polarity of the model
edges is determined based on the image Image. The transformation
that maps the model edges to the image position where the respective object
appears is given in HomMat2D.
The parameter Metric determines the conditions under which the model
is recognized later in the search image. If Metric =
'use_polarity', the object in the image and the model must have the
same contrast. If, for example, the model is a bright object on a dark
background, the object is found only if it is also brighter than the
background. If Metric = 'ignore_global_polarity', the
object is found in the image also if the contrast reverses globally. In the
above example, the object hence is also found if it is darker than the
background. The runtime of find_shape_model will increase slightly
in this case.
It must be ensured that the object edges in the Image have the same
(or inverse) polarity as the object edges in the image in which the object
must be searched later. Especially, the object must not be occluded in the
Image and the background must be either brighter than the object or
darker. Otherwise, the determined polarity of the model edges will not
represent the polarity of the object edges during the search. Note that only
the polarity of the edges is determined, not their contrast. Note also that
the polarity is determined from a single-channel image, only. If a
multichannel image is passed in Image, only the first channel will
be used (and no error message will be returned).
The transformation defined by the matrix HomMat2D maps the model
contours to the respective position where the object appears in the
Image. It can be obtained by finding the object in the
Image, e.g., with the operator find_shape_model and
determining the respective transformation matrix with the operator
vector_angle_to_rigid. The operator set_shape_model_metric
ignores an origin that was set by set_shape_model_origin and always
assumes (0,0) as the origin. Hence, when finding the object in the
Image, all other operators must have the origin at the point (0,0).
Therefore, set_shape_model_origin must be called with (0,0) as the
origin.
A typical proceeding is to read the XLD contours from file. Since these XLD
contours do not provide polarity information, the shape model must be created
from the XLD contours by setting the parameter Metric to
'ignore_local_polarity'. Then, in a first search image the model
is recognized. The transformation that maps the model contours to the
position of the object in the search image can be determined from the
matching result. To verify the match interactively, the model contours can
be mapped to this position. If the matching result is correct, the value of
the parameter Metric can be changed, e.g., to
'use_polarity'. This leads to a faster and more robust recognition
in the following images.
set_shape_model_metric can only be used with shape models that were
created from XLD contours.
This operator modifies the state of the following input parameter:
The value of this parameter may not be shared across multiple threads without external synchronization.
Image (input_object) singlechannelimage → object (byte / uint2)
Input image used for the determination of the polarity.
ModelID (input_control, state is modified) shape_model → (handle)
Handle of the model.
HomMat2D (input_control) hom_mat2d → (real)
Transformation matrix.
Metric (input_control) string → (string)
Match metric.
Default value: 'use_polarity'
List of values: 'ignore_global_polarity', 'use_polarity'
*
* Read the contours from file
read_contour_xld_dxf (Contours, 'metal-part-01', [], [], DxfStatus)
*
* Scale the contours such that they correspond to the image
hom_mat2d_identity (HomMat2DIdentity)
hom_mat2d_scale (HomMat2DIdentity, 3.35, 3.35, 0, 0, HomMat2DScale)
affine_trans_contour_xld (Contours, ContoursAffineTrans, HomMat2DScale)
*
* Create the shape model with Metric='ignore_local_polarity'
create_shape_model_xld (ContoursAffineTrans, 'auto', rad(0), rad(360), \
'auto', 'auto', 'ignore_local_polarity', 5, ModelID)
*
* Get an image of the object and try to find the object in this image
read_image (Image, 'metal-parts/metal-parts-01')
find_shape_model (Image, ModelID, rad(0), rad(360), 0.5, 1, 0.5, \
'least_squares', 0, 0.9, Row, Column, Angle, Score)
*
* Visualize the match
vector_angle_to_rigid (0, 0, 0, Row, Column, Angle, HomMat2D)
get_shape_model_contours (ModelContours, ModelID, 1)
affine_trans_contour_xld (ModelContours, ModelContoursAffineTrans, HomMat2D)
*
* Set the matching metric to 'use_polarity', based on the position
* of the found object
set_shape_model_metric (Image, ModelID, HomMat2D, 'use_polarity')
*
* Use the shape model to detect the object faster and more robust
read_image (Image, 'metal-parts/metal-parts-02')
find_shape_model (Image, ModelID, rad(0), rad(360), 0.5, 0, 0.5, \
'least_squares', 0, 0.9, Row, Column, Angle, Score)
*
* Visualize the matches
for Index := 0 to |Score|-1 by 1
vector_angle_to_rigid (0, 0, 0, Row[Index], Column[Index], Angle[Index], \
HomMat2D)
affine_trans_contour_xld (ModelContours, ModelContoursAffineTrans, \
HomMat2D)
endfor
If the parameters are valid, the operator set_shape_model_metric
returns the value 2 (H_MSG_TRUE). If necessary an exception is raised.
create_shape_model_xld,
create_scaled_shape_model_xld,
create_aniso_shape_model_xld
find_shape_model,
find_scaled_shape_model,
find_aniso_shape_model
create_shape_model_xld,
create_scaled_shape_model_xld,
create_aniso_shape_model_xld
Matching