find_caltab🔗
Short description🔗
find_caltab — Segment the region of a standard calibration plate with rectangularly
arranged marks in the image.
Signature🔗
find_caltab( image Image, out region CalPlate, filename.read CalPlateDescr, integer SizeGauss, integer MarkThresh, integer MinDiamMarks )
Description🔗
find_caltab is used to determine the region of a plane
calibration plate with circular marks in the input image
Image. The region must correspond to a standard calibration plate
with rectangularly arranged marks described in the file
CalPlateDescr. The successfully segmented
region is returned in CalPlate. The operator provides two algorithms.
By setting appropriate integer values in SizeGauss,
MarkThresh, and MinDiamMarks, respectively, you invoke
the standard algorithm. If you pass a tuple of parameter names in
SizeGauss and a corresponding tuple of parameter values in
MarkThresh, or just two empty tuples, respectively, you invoke the
advanced algorithm instead. In this case the value passed in
MinDiamMarks is ignored.
Standard algorithm🔗
First, the input image is smoothed (see gauss_image); the size of
the used filter mask is given by SizeGauss. Afterwards, a
threshold operator (see threshold) with a minimum gray value
MarkThresh is applied. Among the extracted connected regions
the most convex region with an almost correct number of holes (corresponding
to the dark marks of the calibration plate) is selected. Holes with a
diameter smaller than the expected size of the marks
MinDiamMarks are eliminated to reduce the impact of noise.
The number of marks is read from the calibration plate description
file CalPlateDescr. The complete explanation of this file
can be found within the description of gen_caltab.
Advanced algorithm🔗
First, an image pyramid based on Image is built. Starting from the
highest pyramid level, round regions are segmented with a dynamic threshold.
Then, they are associated in groups based on their mutual proximity and
it is evaluated whether they can represent marks of a potential
calibration plate. The search is terminated once the expected number of
marks has been identified in one group. The surrounding lighter area
is returned in CalPlate.
The image pyramid makes the search independent from the size of the image
and the marks. The dynamic threshold makes the algorithm immune to bad or
irregular illumination. Therefore, in general, no parameter is required.
Yet, you can adjust some auxiliary parameters of the advanced algorithm
by passing a list of parameter names (strings) to SizeGauss and
a list of corresponding parameter values to MarkThresh. Currently
the following parameter is supported:
-
'gap_tolerance': Tolerance factor for gaps between the marks. If the marks appear closer to each other than expected, you might set 'gap_tolerance' < 1.0 to avoid disturbing patterns outside the calibration plate to be associated with the calibration plate. This can typically happen if the plate is strongly tilted and positioned in front of a background that exposes mark-like patterns. If the distances between single marks deviate significantly, e.g., if the calibration plate appears with strong perspective distortion in the image, you might set 'gap_tolerance' > 1.0 to enforce the grouping for the more distant marks.
Suggested values: 0.75, 0.9, 1.0, 1.1, 1.2, 1.5
Default: 1.0
Execution information🔗
Execution information
-
Multithreading type: reentrant (runs in parallel with non-exclusive operators).
-
Multithreading scope: global (may be called from any thread).
-
Automatically parallelized on tuple level.
Parameters🔗
Image (input_object) singlechannelimage(-array) → object (byte / uint2)
Input image.
CalPlate (output_object) region → object
Output region.
CalPlateDescr (input_control) filename.read → (string)
File name of the calibration plate description.
Default: 'caltab_100.descr'
List of values: 'caltab_100mm.descr', 'caltab_10mm.descr', 'caltab_200mm.descr', 'caltab_2500um.descr', 'caltab_30mm.descr', 'caltab_650um.descr', 'caltab_6mm.descr', 'caltab_800mm.descr', 'caltab_big.descr', 'caltab_small.descr'
File extension: .descr
SizeGauss (input_control) integer(-array) → (integer / string)
Filter size of the Gaussian.
Default: 3
List of values: 0, 3, 5, 7, 9, 11, 'gap_tolerance'
MarkThresh (input_control) integer(-array) → (integer / real)
Threshold value for mark extraction.
Default: 112
Suggested values: 48, 64, 80, 96, 112, 128, 144, 160, 0.5, 0.9, 1.0, 1.1, 1.5
MinDiamMarks (input_control) integer → (integer)
Expected minimal diameter of the marks on the calibration plate.
Default: 5
Suggested values: 3, 5, 9, 15, 30, 50, 70
Example🔗
(HDevelop)
* Read calibration image.
read_image(Image, 'calib/calib_distorted_01')
* Find calibration pattern.
find_caltab(Image, CalPlate, 'caltab_100mm.descr', 3, 112, 5)
// read calibration image
HImage Image("calib/calib_distorted_01")\;
// find calibration pattern
HRegion CalPlate = Image.FindCaltab("caltab_100mm.descr", 3,112, 5)\;
Result🔗
find_caltab returns 2 (H_MSG_TRUE) if all parameter values are
correct and an image region is found. The behavior in case
of empty input (no image given) can be set via
set_system(::'no_object_result',<Result>:)
and the behavior in case of an empty result region via
set_system(::'store_empty_region',<'true'/'false'>:).
If necessary, an exception is raised.
Combinations with other operators🔗
Combinations
Possible predecessors
Possible successors
See also
find_marks_and_pose, camera_calibration, disp_caltab, sim_caltab, caltab_points, gen_caltab
Module🔗
Foundation