[ITK-dev] [ITK] Itk release branch demons registration

Bradley Lowekamp brad at lowekamp.net
Sun Jan 17 18:13:44 EST 2016 

Just to follow up my git bisect narrowed the change down to this:
commit 0fd462a3226e84bdce77b68830903b3888dab244
Author: Hyun Jae Kang <hyunjae.kang at kitware.com>
Date:   Thu Dec 31 10:56:25 2015 -0500

    BUG: Fixed the runtime crash of ITKStatisticsTestDriver tests on OSX 10.6
    
    - Fixed the runtime crash of the following ITKStatisticsTestDriver's tests on
      OSX 10.6 ( clang 3.0 )  by modifying the itkHistogram's Initialize function
      to handle a data value out of range of a specific data type with
      type-casting operator.
    
      - itkSampleToHistogramFilterTest3 (ILLEGAL)
      - itkSampleToHistogramFilterTest7 (ILLEGAL)
    
    - These crash were reported at the following page:
      https://open.cdash.org/viewTest.php?onlyfailed&buildid=4170483
    
    Change-Id: I2063912edee79422d88125ffd5f7513c31a9e98c

diff --git a/Modules/Numerics/Statistics/include/itkHistogram.hxx b/Modules/Numerics/Statistics/include/itkHistogram.hxx
index 380b8d0..d11d788 100644
--- a/Modules/Numerics/Statistics/include/itkHistogram.hxx
+++ b/Modules/Numerics/Statistics/include/itkHistogram.hxx
@@ -248,8 +248,8 @@ Histogram< TMeasurement, TFrequencyContainer >
     {
     if ( size[i] > 0 )
       {
-      interval = static_cast<float>( upperBound[i] - lowerBound[i] )
-        / static_cast< MeasurementType >( size[i] );
+      interval = (static_cast<float>( upperBound[i] ) - static_cast<float>(lowerBound[i] ))
+        / static_cast< float >( size[i] );


Here is the failing example [1], which uses the HistogramMatching filter.

I am not sure why float was choose at the type here. Are results being truncated now? When there was temporary double computation going on?

 It needs more investigation, as it’s changing the results of some filters… not sure how significant yet.

Brad




[1] https://github.com/SimpleITK/SimpleITK/blob/master/Examples/DemonsRegistration1.py#L41
> On Jan 16, 2016, at 2:58 PM, Bradley Lowekamp <brad at lowekamp.net> wrote:
> 
> Thanks. I'll run a git bisect tonight to try to figure it out. 
> 
>> On Jan 16, 2016, at 12:57 PM, Matt McCormick <matt.mccormick at kitware.com> wrote:
>> 
>> Hi,
>> 
>> Nothing comes to mind.
>> 
>> Here is the log for v4.9rc03..release:
>> 
>> Bill Hoffman (3):
>>     COMP: fix 64 bit build warnings with windows auto-export on.
>>     COMP: work around for VS 2015 optimizer bug causing test failures.
>>     COMP: work around for VS 2015 optimizer bug causing test failures.
>> 
>> Bradley Lowekamp (1):
>>     BUG: Fix precision with accumulation and scaling in AdaptiveHistogram
>> 
>> Davis Vigneault (1):
>>     COMP: Include itkMacro.h in itkTestingMacros.h
>> 
>> Hans Johnson (2):
>>     COMP: BSD command lines do not have --version
>>     COMP: Respect CMAKE_CXX_STANDARD during config
>> 
>> Hyun Jae Kang (12):
>>     COMP: Fixed the compiler error of ITKCommon2TestDriver on OSX 10.6
>>     BUG: Fixed the runtime crash of vnl_test_complex on OSX 10.6
>>     BUG: Fixed the runtime crash of itkTimeProbeTest2
>>     BUG: Fixed the runtime crash of VideoSourceTest on OSX 10.6
>>     BUG: Fixed the runtime crash of ITKReviewTestDriver on OSX 10.6
>>     BUG: Fixed the runtime crash of ITKFastMarchingTestDriver's
>> tests on OSX 10.6
>>     COMP: Fixed the data conversion warning messages of itkResourceProbe
>>     BUG: Fixed the runtime crash of ITKStatisticsTestDriver tests on OSX 10.6
>>     BUG: Fixed the runtime crash of itkBinaryShapeOpeningImageFilterTest1
>>     BUG: Fixed the runtime crash of test_pow_log on OSX 10.6
>>     BUG: Fixed the runtime crash of vnl_test_numeric_traits
>>     BUG: Exclude a test code of ITKLabelMapTestDriver on OSX 10.6
>> 
>> Isaiah Norton (1):
>>     COMP: fix build with -std=c++11 when ITK/VXL initially configured without
>> 
>> Matthew McCormick (6):
>>     BUG: Do not perform dynamic_cast in CompositeIOTransformIOHelper.
>>     COMP: Do not set property on itkhdf5 with ITK_USE_SYSTEM_HDF5.
>>     DOC: Update the location of the GNUPlot Software Guide scripts.
>>     DOC: Update the Software Guide repo location in Examples/README.txt.
>>     COMP: Add export specification for itk::ResourceProbe.
>>     COMP: Require NO_MODULE with DCMTK find_package.
>> 
>> Michka Popoff (2):
>>     ENH: Allow Python 3 wrapping for Glue and Review Modules with VTK 7
>>     COMP: Re-enable review module for python 3 and older VTK's
>> 
>> Sean McBride (1):
>>     DOC: Update some woefully out-of-date GDCM comments, and typos
>> 
>> Seun Odutola (1):
>>     BUG: Improved itkAnalyzeImageIO to handle case insensitive extensions
>> 
>> HTH,
>> Matt
>> 
>>> On Sat, Jan 16, 2016 at 12:18 PM, Bradley Lowekamp <brad at lowekamp.net> wrote:
>>> I just updated SimpleITK to the release branch yesterday and there are some changes in the result of the demons registration. I'm not going to get a chance to look at it until Monday. The only change potentially related I know of is the rounding related numeric traits, but that does not seem likely to me.
>>> 
>>> Any ideas why the Demons Registration would have changed fro rc3?
>>> 
>>>> On Jan 16, 2016, at 12:13 PM, Bradley Lowekamp <brad at lowekamp.net> wrote:
>>>> 
>>>> Hello,
>>>> 
>>>> I updated SimpleITK
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>> 
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html
>>>> 
>>>> Kitware offers ITK Training Courses, for more information visit:
>>>> http://kitware.com/products/protraining.php
>>>> 
>>>> Please keep messages on-topic and check the ITK FAQ at:
>>>> http://www.itk.org/Wiki/ITK_FAQ
>>>> 
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://public.kitware.com/mailman/listinfo/insight-developers
>>> _______________________________________________
>>> Powered by www.kitware.com
>>> 
>>> Visit other Kitware open-source projects at
>>> http://www.kitware.com/opensource/opensource.html
>>> 
>>> Kitware offers ITK Training Courses, for more information visit:
>>> http://kitware.com/products/protraining.php
>>> 
>>> Please keep messages on-topic and check the ITK FAQ at:
>>> http://www.itk.org/Wiki/ITK_FAQ
>>> 
>>> Follow this link to subscribe/unsubscribe:
>>> http://public.kitware.com/mailman/listinfo/insight-developers
> _______________________________________________
> Powered by www.kitware.com
> 
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
> 
> Kitware offers ITK Training Courses, for more information visit:
> http://kitware.com/products/protraining.php
> 
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
> 
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/insight-developers
> _______________________________________________
> Community mailing list
> Community at itk.org
> http://public.kitware.com/mailman/listinfo/community

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/insight-developers/attachments/20160117/188815c1/attachment.html>

More information about the Insight-developers mailing list