[Insight-users] bug in Morphology with parabolic structuring elements?

Richard Beare richard.beare at gmail.com
Thu Sep 16 19:12:34 EDT 2010


Hi,

The bright edge was definitely a bug - I forgot to change the way the
"SafeBorder" was implemented for a closing. I think that is fixed now.
Note that if your image already has a natural boundary then you can
turn this off with the SetSafeBorder

The streaks that you're seeing that aren't related to the border are
correct, I think. That sort of thing happens if the image structure
matches the scale of the structuring element.

The blank image problem is less obvious, but I can think of a couple
of things to check.

Does the input image have voxel spacing information? If so, does the
scale of the structuring element seem appropriate. If the scale was
very large then the bug with the edge safety would lead to an image
with every voxel set to the maximum.

Anyway, the changes are committed, so we can test it out.

PS dilateBinary had been moved to dilateBinary3D

On Fri, Sep 17, 2010 at 1:48 AM, Oleksandr Dzyubak <adzyubak at gmail.com> wrote:
>
> Hi Richard,
>
> Many thanks for you prompt response.
> I successfully downloaded the latest version from the link you provided.
> I see that you greatly updated your package. Now you have 47 files vs 31 in
> the submission!
> In your Cmake file there was a typo giving me the error below.
>
>  CMake Error in CMakeLists.txt:
>   Cannot find source file "dilateBinary.cxx".  Tried extensions .c .C .c++
>   .cc .cpp .cxx .m .M .mm .h .hh .h++ .hm .hpp .hxx .in .txx
>
> So to get rolling, I just commented out the "dilateBinary.cxx" file from my
> building list.
>
> Yes, you are right. Now testClose works but in a way I did not expect it to
> (see uploaded result).
>
> Closed cthead1 Image.
> http://old.nabble.com/file/p29729933/close_cthead1_x1_y05.png
> close_cthead1_x1_y05.png
>
> I observe couple "artifacts" there (which do not appear in testOpen! ):
> "White Border" and "Severe Streaks".
>
> However real Mystery starts later when I try to apply your Open-Close
> algorithms to my images.
> To demonstrate the problem, I also uploaded the input, opened, and closed
> images.
>
> Input Image.
> http://old.nabble.com/file/p29729933/SheppLogan_Proj_720.jpg
> SheppLogan_Proj_720.jpg
>
> Opened Image.
> http://old.nabble.com/file/p29729933/open_SheppLogan_Proj_720.jpg
> open_SheppLogan_Proj_720.jpg
>
> Closed Image.
> http://old.nabble.com/file/p29729933/close_SheppLogan_Proj_720.jpg
> close_SheppLogan_Proj_720.jpg
>
> As you can see, testOpen still works but testClose does not.
> Is that what you also observe in your tests or it is just me?
>
> Regards,
>
> Alex
>
>
> Richard Beare wrote:
>>
>> Hi,
>> There have been a few updates that aren't in the IJ version. You can
>> get the latest version using
>>
>> darcs get http://voxel.jouy.inra.fr/darcs/contrib-itk/parabolicMorphology
>>
>> The biggest changes are classes for fast circular and spherical
>> erosions and dilations on binary images.
>>
>>
>> I've added testClose, in the way that you described, and it seems to
>> run sensibly on the cthead test image. It is possible that there is a
>> bug in the IJ version that I've fixed since.
>>
>>
>> On Thu, Sep 16, 2010 at 6:06 AM, Oleksandr Dzyubak <adzyubak at gmail.com>
>> wrote:
>>>
>>> Dear ITK users,
>>>
>>> Recently I tried classes "Morphology with parabolic structuring elements"
>>> by
>>> Richard Beare.
>>> Since it is not available in the ITK lib yet,
>>> I downloaded the submission version from the link below.
>>>
>>> http://www.insight-journal.org/browse/publication/228
>>>
>>> BTW, on the submission page there is a recommendation which basically
>>> does
>>> not work.
>>> *************
>>> Please use this identifier to cite or link to this publication:
>>> http://hdl.handle.net/1926/1370
>>> *******
>>> If I click on the provided link,
>>> it gives (me at least) the error message "You cannot access this item!"
>>>
>>> In submission everything compiles well and provided examples work.
>>> I am, in particular, interested in "Open-Close" morphological operations.
>>> Richard provided an example testOpen.cxx which works
>>> and produces reasonable results.
>>> Thus, I took it as basis to build testClose.cxx.
>>>
>>> To make my life easier, I included
>>> #include "itkParabolicCloseImageFilter.h"
>>>
>>> and substituted
>>> typedef itk::ParabolicOpenImageFilter< IType,IType > FilterType;
>>> with
>>>  typedef itk::ParabolicCloseImageFilter< IType,IType > FilterType;
>>>
>>> After those manipulations, I was able to succesfully compile my code
>>> testClose.cxx.
>>> However, as opposed to testOpen.cxx code,
>>> in the output image all the pixels have the same value 3.4E38.
>>>
>>> Is this a bug or I am doing something terribly wrong?
>>>
>>>
>>> Thanks,
>>>
>>> Alex
>>>
>>> _____________________________________
>>> 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://www.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-users
>>>
>> _____________________________________
>> 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://www.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-users
>>
>>
>
> --
> View this message in context: http://old.nabble.com/bug-in-Morphology-with-parabolic-structuring-elements--tp29722504p29729933.html
> Sent from the ITK - Users mailing list archive at Nabble.com.
>
> _____________________________________
> 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://www.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-users
>


More information about the Insight-users mailing list