Page MenuHomePhabricator

Crash in Rigid Registration - Point Based Registration upon Register
Closed, ResolvedPublic

Description

Workflow

  • Load all the two test images and point sets for point based registration
  • select the two images
  • press Register

-> Crash

Event Timeline

did you press the register button in the RigidRegistration view?

I tried that and it does not crash on my computer.

Did you load a specific preset?

I can observe, that the registration uses a lot of memory. The MITK3M3.exe uses >3GB Ram, which would certainly throw a out-of-memory exception on a 32 Bit machine. This could be the source of the crash

another try with the 32 Bit installer:

The combination

fixed: venous image
moving: arterial image
just does nothing. If I click on register, nothing happens, no error message.

Next try:

If I switch the fixed and moving images, the algorithm starts, the progress bar is slowly filling.

The log output (at 13% completion):
[333.48] Node 'LeberArteriell' selected
[333.50] Reference image 'LeberArteriell' selected
[333.50] Node 'LeberVenoes' selected
!333.50! WARNING: WARNING: No image or too many (>2) were selected.
[333.51] Reference 00000000 Working 00000000
[333.51] Updating alignment warning
[333.53] Nothing selected, re-show all segmentations
Ignore this selection event: sourcepart == this 0 selection == NULL1
Moving Image Geometry (IndexToWorldTransform)
0.61 0.00 0.00
0.00 0.61 0.00
0.00 0.00 2.00
center 0.00 0.00 0.00
offset -155.20 -290.42 710.00

RigidRegistrationParameters.xml is empty or does not exist. There are no presets
to select.
[562.43] 135445 [-9, -10, -10]

Translation is: -9.00-10.00-10.00

There is no RigidRegistrationParameters.xml in the install directory, but a mitkRigidRegistrationPresets.xml! Looking at the source code, it seems that it is trying to load the correct file, just the error message contains the wrong name.

I waited for ~15 minutes, the rigid registration was still running (at 20%). It did not crash and used ~940MB RAM.

Many of the logged messages seem to come from the segmentation module which was most probably active at the same time...

I think the first combination has printed the error message for failed to allocate memory to the console, maybe this can be changed to an message box?

(In reply to comment #4)

Many of the logged messages seem to come from the segmentation module which was
most probably active at the same time...

I think the first combination has printed the error message for failed to
allocate memory to the console, maybe this can be changed to an message box?

I changed the log message to a message box. See T3060.

I tried again with x64 build, revision #20542. It does not crash anymore, the segmentation seems to be working correctly.

installer x86 version, revision #20550: A "failed to allocate memory" messagebox appears, the application does not crash.

Problem seems to be fixed as much as possible on 32 bit systems... Closing bug now.

Merging "applicazion modules" component with "ExtApp plugins"