[Insight-developers] ScalarChanAndVese LevelSet tests failing

[Arnaud Gelas]

Bill,

Thanks a lot for tracking and fixing these errors!!!

Unfortunately, I am not able to track the remaining error from here :-(...

I have not spend time on the ScalarChanVese****ImageFilter yet, but I 
will try to look at it this afternoon.

Arnaud

Bill Lorensen wrote:
> Arnaud,
>
> The Borland build was reporting numerical errors on the two
> ChanAndVese level set tests. I tracked it down to the RIO filter used
> in the Initialize() method of the two filters. The size of the region
> was (0,0). This is because the image does not exist when the
> SetLevelSet method is called in the test. The reader that provides the
> level set is never updated.
>
> I fixed the tests by updating the reader before apply the SetLevelSet
> method, but this is not the best fix. The filter should make sure
> these images are up-to-date.
>
> However, the Sparse test is still failing with a floating point
> overflow. CalculateUpdateValues is returning a large 1.e23 value. The
> change argument is 1.e24. That's about all the help I can provide.
>
> Also, the ADD_TEST commands were a bit messed up.
>
> Bill
>