[Insight-users] Re: help for BinaryThinningImageFilter

[Luis Ibanez]

Hi Qxue,


Both the result of the ITK and the ImageJ filters seems to be
correct, depending on how you want to treat the boundary.

The ITK output is correct, if you assume that the dark boundary
of the image continues to infinity.

The ImageJ output is correct, if you assume that the intensity
values outside of the image are white.

ImageJ is giving you skeleton lines in the middle distance between
the cells and the image boundary.

I prefer the output of the ITK filter, since it doesn't introduce
artificial boundaries.

In the case of your application, you should decide what type of
boundary condition makes more sense.



     Regards,


        Luis



-------------------------
qxue at jouy.inra.fr wrote:
> Hi Luis:
> 
>    I have sent an email to th ITK list, but I want to sent it to you again with
> the attachments for a detailed description about my question.
>    I meet with a question using the BinaryThinningImageFilter, for this filter
> can not get the correct result in the edge of the input image when there are no
> edge to compute.
>    As a result the skeleton of the image is to push the one-pixel-wide edge to
> the image's edge directly.
>    ImageJ can get a closed shape on the edge of the image.
>    Is it a bug in this filter or am I wrong to use it?
> 
> Quan
> 
> 
> 
> 
> 
> 
> ------------------------------------------------------------------------
> 
> 
> ------------------------------------------------------------------------
> 
> 
> ------------------------------------------------------------------------
>