[Insight-developers] PolyLineParametricPath

[Lorensen, William E (GE, Research)]

I agree that replacing a bad algorithm with a better one should be a bug fix and not a new class. One caution is that the API should not change.

Bill


-----Original Message-----
From: insight-developers-bounces+lorensen=crd.ge.com at itk.org
[mailto:insight-developers-bounces+lorensen=crd.ge.com at itk.org]On Behalf
Of Zachary Pincus
Sent: Monday, January 30, 2006 3:09 PM
To: Luis Ibanez
Cc: Jeffrey Duda; Insight Developers List
Subject: Re: [Insight-developers] PolyLineParametricPath 


Hi folks -

Yesterday I wrote up a 'marching squares' implementation to extract  
image contours into a PolyLineParametricPath, so this discussion is  
vaguely relevant to me. (The plan is to submit that to the Insight  
Journal so that ITK can have a proper 2D contour extractor filter.)

Given that, my (minor) preference is to have this fix incorporated  
(via Journal submission or no) into  
PolyLineParametricPath::IncrementInput() instead of creating a  
separate subclass. Having to decide which subclass to output is a  
pain for people writing path creation filters.

Perhaps Jeffrey can clarify this: are there any cases in which the  
current ParametricPath::IncrementInput() would be superior to the  
proposed replacement for PolyLineParametricPath?

If not, I see no reason not to include the patch directly into the  
PolyLineParametricPath class. If so, that would argue for making a  
subclass, or figuring out how to automatically choose which algorithm  
to use, or providing an intelligent default and having a user- 
selectable switch. In my mind, the latter two would be better than  
subclassing.


Zach Pincus
Program in Biomedical Informatics and Department of Biochemistry
Stanford University School of Medicine



On Jan 30, 2006, at 11:21 AM, Luis Ibanez wrote:

>
> Hi Jeffrey,
>
> Thanks for pointing out the problem with
> the PolyLineParametricPath class.
>
> I will be great to fix the class in the CVS repository,
> but if you think that your alternate class is a better
> choice, we are also open to add your new class to the
> toolkit.
>
> Do you have in mind a particular place of the code
> where we could fix the PolyLineParametricPath ?
>
>
>
>
> If you would like to contribute your class to ITK,
> the mechanism will involve to write a paper to the
> Insight Journal. Note that these papers are closer to
> technical reports than to traditional journal papers.
>
> Instructions on how to prepare a paper for the
> Insight Journal are available at:
>
> http://www.itk.org/Wiki/ 
> ITK_Procedure_for_Contributing_New_Classes_and_Algorithms
>
> A template for preparing a submission is available at:
>
> http://www.na-mic.org:8000/svn/NAMICSandBox/trunk/InsightJournal/ 
> SubmissionTemplate/
>
> and as a tar.gz file at:
>
> ftp://public.kitware.com/pub/itk/InsightJournal/ 
> InsightJournalSubmissionTemplate.tgz
>
>
> Please let us know what would be your preference,
>
>
>    Thanks
>
>
>      Luis
>
>
>
> -------------------
> Jeffrey Duda wrote:
>> Luis,
> ...snip
>
> ... I wrote an email to the list a while ago (but
>> got not response) about some problems with the  
>> PolyLineParametricPath and the fact that the algorithm used to  
>> step along the PolyLine to find the next voxel in the Path is  
>> unstable (especially in data with anisotropic voxel spacing).  It  
>> may be that I am the only who uses it so the issue hasn't bothered  
>> anyone else.  Either way, it may be worth either a fix or at least  
>> a note in the documentation.  I wrote a very basic alternate class  
>> ( http://www.seas.upenn.edu/~jtduda/itkother.html#slowline ).  I'd  
>> be happy to work on a modification to the existing class if you  
>> think that would be useful/beneficial.
>
> .. snip...
>> thanks a lot for your help
>> -jeffrey duda
>
> _______________________________________________
> Insight-developers mailing list
> Insight-developers at itk.org
> http://www.itk.org/mailman/listinfo/insight-developers

_______________________________________________
Insight-developers mailing list
Insight-developers at itk.org
http://www.itk.org/mailman/listinfo/insight-developers