[Insight-developers] itkSimpleImageRegistrationTest test failings

[M Stauffer -V-]

Hi Matt, 

Yes, I see. I was focusing on the other failing builds, but maybe the
div-by-zero is changing results those somehow differently than in the
passing builds.

I'm merging Gang's fix to the div-by-zero in ANTSCC metric, and will
retry itkSimpleImageRegistrationTest after that, possibly with the
tighter convergence criteria.

-M

>Increasing the tolerance appears to cover up a real problems.
>
>A preferable path would fix the divide-by-zero error pointed 
>out by Sean.
>
>Then, it may help change the convergence criteria on the the optimizer
>to specify a tighter convergence of the affine translation.
>
>Thanks,
>Matt
>
>On Tue, Jan 31, 2012 at 5:50 PM, M Stauffer -V- 
><mstauff at verizon.net> wrote:
>> Hi,
>>
>> Regarding the failing itkSimpleImageRegistrationTest tests here:
>>
>> 
>http://www.cdash.org/CDash/testSummary.php?project=2&name=itkSi
mpleImageRegistrationTest&date=2012-01-30
>>
>> In patch  http://review.source.kitware.com/#/c/3693/ I 
>updated the baseline
>> image for the test
>> 
>(Modules/Registration/RegistrationMethodsv4/test/Baseline/itkSi
>mpleImageRegistrationTest.nii.md5).
>>
>> Is it possible that the image didn't get updated in the 
>builds for which the
>> test is failing? 3 out of the 4 failing tests report the 
>exact same number
>> of failing pixels. The std output of the failing tests show 
>some state
>> values different than the passing tests, however there is a 
>passing test
>> with virtually identical state values as the failing tests.
>>
>> In the meantime, I'm putting up a patch that adds
>> "--compareNumberOfPixelsTolerance 15000" to the test which 
>should cover
>> things if I understand things correctly:
>> http://review.source.kitware.com/#/c/3786/
>>
>> -M
>>
>> _______________________________________________
>> 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.html
>>
>> 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://www.itk.org/mailman/listinfo/insight-developers
>>