screenshot2

screenshot1

I have another image (standard direction) with spacing of 40, 40, 48 and also get strange results. See screenshot3. Here, the manually segmented layer is not fully included in the interpolation.

screenshot3

Current release is finished. Resetting target milestone

Guys,

This is a major issue for us. The segmentation module is unusable with 2014.10.
-F

I will have a look at that for the upcoming release.
@Felix: are there other issues regarding the segmentation plugin? You wrote that it is unusable with 2014.10. Or was did you mean that the 3d interpolation is unusable?

User fetzer has pushed new remote branch:

bug-17434-3d-interpol-spacing-issue

I fixed the strange results for the images with a very small spacing.
I could not reproduce the bug where the manual segmentation is not fully included in the interpolation.

There is still an issue for images with bigger spacing. It seems that the distance values are calculated for the whole image instead of just along a narrow band. I will investigate that

[075ce8]: Merge branch 'bug-17434-3d-interpol-spacing-issue-v2'

Merged commits:

2015-03-30 11:48:38 Andreas Fetzer [1b0370]
Updated MITK-Data and reactivated test for windows

2015-03-30 11:39:05 Andreas Fetzer [6b3b76]
Make buffer value a member variable

2015-03-30 11:38:43 Andreas Fetzer [a78ecb]
Cosmetic changes

2015-03-30 11:37:56 Andreas Fetzer [69147b]
Replaced deprecated function call

2015-03-30 10:56:51 Andreas Fetzer [c5ff44]
Fixed line width of surface interpolation feedback in 2D

2015-03-25 19:01:58 Andreas Fetzer [c1a373]
Fixed spacing calculation for distance image

Now the spacing can have a float value which is needed in case of reference data with a very small spacing

2015-03-30 11:02:33 Andreas Fetzer [765a16]
Fixed the calculation of the off-surface points

Now the normals used for the calculation are scaled with the spacing of the
distance image.
Previously the normals always had the size of 1 mm, which is too small or too big
in case of images with a very big or small spacing

2015-03-25 14:52:53 Andreas Fetzer [58859e]
Fixed normal calculation.

The normals were not normalized properly.

2015-03-18 22:59:25 Andreas Fetzer [f981c2]
Fixed weird results for images with a very small spacing

The reason was that the narrow band calculation of the distance image values
used a too small threshold.

2015-03-18 15:11:42 Andreas Fetzer [17b99b]
COMP Merge branch 'bug-18603-comp-fix'

2015-03-18 15:09:16 Andreas Fetzer [ffe274]
Merge branch 'master' of mitk.org:MITK

[8453d6]: Merge branch 'bug-17434-3d-interpol-spacing-issue-v2'

Merged commits:

2015-03-30 15:58:08 Andreas Fetzer [34138c]
COMP Fixed 32 bit test failure which occurred due to some rounding error in the
itk transform helper classes.

Now we are avoiding this by using a continuous index