The pushed branch contains a temporary fix for the problem: instead of using mitk::eps, the image raster alignment and image orientation now must be smaller than 0.001.

(In reply to comment #1)

Andi, I don't know that feature, but I guess this mechanism is not included
in the image statistics bundle, at least I can't remember any code for that
feature. Mathias, did you implement that feature in the image statistics
bundle?

REVIEW of the solution proposed in bug-12894-fix-global-tnc-handling (see wiki specification page):

REVIEW: The implemented solution removes the hover-effect of PlanarFigure control points after clicking on a control point. When the mouse is moved outside of the control point area, no false hover effect is shown anymore.

Fix seems reasonable; the only possibly annoying issue that we noticed is that when you double-click onto another control-point of the same polygon (placed somewhere completely else), polygon creation is also finished, and this last position of the mouse is ignored. This seems to be acceptable though, because it is highly unlikely that users interact with the polygon like this.

[ca0cf0]: Merge branch 'bug-9672-smooth-roi-not-finalized-correctly'

Markus fixed this bug:

Problem solved and solution commited to master and to branch releases/snapshot-201109.

[43739f]: Merge branch 'bug-9232-planarfigure-control-point-overlap'

[c943b9]: Merge branch 'bug-8959-planar-figures-not-updated'

Many thanks for sending this fix! Someone will look into how this can be integrated into MITK

(In reply to comment #4)

Can anyone explain that to me (Mathias perhaps :-))?

[SVN revision 29373]
ENH (#6737): add state-machines for right mouse button zooming and panning via ctrl / shift modifiers

[SVN revision 29345]
FIX (#6729): ensure correct out-of-bounds checking for planar figures in extrema situations

Problem behoben: schuld war der Index-Offset von -0.5/-0.5/-0.5, der NACH dem out-of-bounds-Check auf die Punkte des Polygons addiert wurde, sodass es in Randfällen (x- bzw. y-Koordinate < 0.5) unerlaubterweise zu Werten außerhalb der Image-Bounds kommen konnte. Der Check wird jetzt erst nach dem AUfaddieren durchgeführt, was das Problem beseitigt.

i have no idea how to solve this --> reassign

[SVN revision 27453]
COMP (#5999): remove incorrect typename

Image masks smaller than input image are now allowed. Several checks have been added to ensure that image and mask have same spacing and that voxels are aligned (i.e. it is clear which mask voxel belongs to image which, no significant overlap between voxels).

[SVN revision 27441]
ENH (#5999): enable mitk::ImageStatisticsCalculator to deal with image masks smaller than the original gray-value image

working

[SVN revision 26437]
ENH (#5528): Do not use anti-aliasing when painting control points as squares

[SVN revision 26387]
ENH (#5528): Make PlanarFigure control points appearance nicer

Geschätzer Aufwand

[SVN revision 26276]
ENH (#5489): Added "iseditable" property to planar figures; if set to false, editing will not be possible

Done.

Done.

Note: changes in StateMachine.xml files committed with r26251

GEschätzer Aufwand

[SVN revision 26275]
ENH (#5424): Add documentation for planar figure properties

[SVN revision 26274]
ENH (#5424): Extend color scheme for planar figure drawing; some code-cleanup and small style changes how PFs are drawn

Geschätzer Aufwand

[SVN revision 26250]
ENH (#5420): Add ability to pick PlanarFigure objects in render windows

• Picking mechanism defined in PlanarFigureInteractor and StateMachines
• Added select-event to PlanarFigureInteractor
• Adapted PlanarFigureMapper2D to highlight figures over which the mouse is hovering
• Adapted QmitkMeasurement to listen to PlanarFigure select-events

[SVN revision 26144]
ENH (#5373): Use custom events for planar figure interaction instead of default ITK events for flexibility

Closing bug.

Do you have a particular dataset on which this problem occured? Was it the only dataset that was displayed, or were there other data on the screen? Does the squeezed image always stay like this, or does e.g. a geometry re-init on the dataset resolve the problem?

(In reply to comment #1)

I'm not able to reproduce the error. I tried it using the measurments tool and
there it works correctly. Can you describe a precise scenaria in which it
fails?

Not yet fixed --> reassign to default

(In reply to comment #2)

I am quite sure that my absdiff filter works correctly so there is no value =
-1023!!!

Any suggestions?

Currently any pixel values that are smaller than -1023 are rendered transparent. This is done to ensure that when multiple images are displayed on top of each other, smaller images do not cover larger images. This would otherwise be the case because the area outside of an image, but inside of the rendered frame (defined by the current WorldGeometry), also is resampled (and assigned a gray value of -1024 by default) and thus displayed.

May actually be related to bug / feature request #5092

no, that's a party bug that anybody could look into

(In reply to comment #6)

(In reply to comment #4)

In addition there should be a test to check the boundaries against the control
points, so that other test cases won't fail because of the same problem.

Note: the reason for the precision loss is probably related to the tiny xml serializer. When writing real-type values in XML text format and reading them in again, some of the prevision is lost. Maybe tiny xml can be configured to write real-type values with higher precision.

Thanks for looking into this. I also had another look at the affected classes. All the MITK Geometry classes use "float" instead of "double" for defining ITK transforms, vectors etc. You're right that this is probably the reason for the precision loss.

[SVN revision 23124]
ENH (#4126): Added constraint to PlanarCross so that the two lines always intersect; extended test case.

[SVN revision 23638]
COMP (#4126): Make really sure that the lines of PlanarCross cross each other

If I remember correclty code block is necessary for the case when an image mask (binary image) or planar figure is selected as a CHILD element of a gray value image. Statistics is then computed over the region of the image defined by the mask or planar figure.

Adding additional worked hours

[SVN revision 22868]
COMP (#4006): Added missing org.mitk.planarfigure bundle and PlanarFigureObjectFactory

A new module "PlanarFigure" exists with r22866. Besides moving all PlanarFigure related classes from MitkExt into this module, a few other measures were necessary:

[SVN revision 22869]
COMP (#4006): Added missing dependency on module PlanarFigure

[SVN revision 22870]
ENH (#4006): Improve order of module includes for better build configuration behavior

[SVN revision 22866]
ENH (#4006): Relocate PlanarFigure classes into separate module; some necessary restructuring

Adding worked hours for fixing this issue and creating a test

[SVN revision 22875]
ENH (#4003): Change way how properties of PlanarFigures (such as open/closed, single/double line, etc.), are stored (now as MITK properties)

[SVN revision 22880]
ENH (#4003): Add unit test for PlanarFigure serialization and deserialization

[SVN revision 22879]
FIX (#4003): Deserialize PlanarFigure spacing

Depends on T4006 (creation of PlanarFigure module)

With the changes in the above commit, the required geometry information (contained in mitk::AffineGeometryFrame3D) is serialized. This consists of ITK transform parameters, plane bounds, spacing and origin.

[SVN revision 22756]
FIX (#3984): Extend PlanarFigureReader/Writer with serialization of entire Geometry information; add PlanarCross to readable PlanarFigure classes

[SVN revision 23214]
COMP (#3965): Adapt test to reduced accuracy for parallelism check

I just searched through all .cpp and .h files of MITK to check the usage of the affected mitk::PlaneGeometry methods.

The following definition of parallelity is currently used in PlanarFigureMapper2D as a workaround, maybe this helps:

[SVN revision 23212]
FIX (#3965): Reduce accuracy when checking for parallelism

[SVN revision 22518]
FIX (#3834): Fix PlanarFigureWriter to write up/right vectors in index and not in world coordinates

[SVN revision 22436]
FIX (#3786): Generate helper line before checking if it must be drawn

Marking bug as party.

[SVN revision 22758]
ENH (#3619): Added ResetFigure to PlanarFigure state-machine to allow modifications after figure placement

[SVN revision 22791]
ENH (#3619): Added alternative mode for drawing lines by mouse-drag (instead of mouse-clicks) to PlanarFigure state-machine

[SVN revision 22789]
ENH (#3619): Added alternative mode for drawing lines by mouse-drag (instead of mouse-clicks) to PlanarFigure state-machine

[SVN revision 22546]
DOC (#3619): Added class documentation

[SVN revision 22757]
ENH (#3619): Add PlanarCross to list of serializable PlanarFigure classes