[Insight-developers] Re: Remaining Failing Tests

[Luis Ibanez]

Bill,


I just committed a fix for both tests.


1) It was checking for 1e-5 in the precision
    of the central pixel. Now it is using a
    percent.


2) This one is funny...

    This is an example on the use of MeanSquares metric,
    but the CMakeLists.txt file was providing two images
    of different modalities (T1 and PD).

    It is surprising that the test was working before...

    Furthermore, it still works for these to multi-modality
    images if we push the initial step to 40 instead of 4,..
    but this is just a coincidence due to the similarity
    between graylevels in T1 and PD.



       Luis




---------------------------------------
Lorensen, William E (Research) wrote:

> I checked in new baselines for some regression tests.
> 
> There are two remaining failing tests.
> 
> 1) itkRecursiveGaussianImageFilterTest - this seems to check for valid
> normailzation of the gaussian across scales. It is failing its internal
> check for consistency.
> 
> 2) ImageRegistration3Test - I did not check in a new baseline image for this
> since it contains in it's comments the output of the registration
> iterations.
> 
> Bill
> 
>