[Insight-developers] MultiScaleHessianBasedMeasureImageFilter in Review...

[Luca Antiga]

Hi Bradley,
  I'm responsible for (part of) the filter, which I modified from the  
first VED IJ paper.
There are a few changes that are already planned for the class from  
discussions with Andinet (who did the migration to the Review  
directory) and Luis. They should be committed in the near future.
To answer your questions:
1) this was made to limit memory consumption in cases where one is not  
interested in the scales or Hessian image. Now, I probably did it the  
wrong way, but I think there is value in letting the user decide how  
much information to get from the filter. Do you have suggestions on  
how to do it right?
2) agreed, feel free to switch to the Buffered region.
3) the update buffer is not redundant in case the OutputPixelType is  
not double (could be made float, to save memory, no real advantage  
here in using doubles).
Thanks for looking into this, looking forward to hearing your  
suggestions

Luca


On Mar 10, 2009, at 3:56 PM, Bradley Lowekamp wrote:

> Hello,
>
> 	I just committed some changes for bug 8681, which will cause some  
> undesired behavior for this class, due to underlying bugs in it.
>
> 	I am not entirely sure of what the guidelines are for editing  
> things in the Review directory are. I have looked at this class and  
> there are some straight forward changes to fix the bug I revealed.  
> But there are some ways this filter does things that doesn't uses  
> best practices. Can these be changed?
>
> 1) Also this class uses optional outputs. That is there is a boolean  
> member variable which indicates if the outputs should be allocated  
> and written. I can't recall seeing this before in a filter. What is  
> weird about this is that this doesn't make much since with the  
> NumberOfRequiredOutpus. I don't think that this is well supported in  
> the pipeline.
>
> 2) This filter uses GetLargestPossibleRegion instead of Buffered  
> region. Interesting things may occur if a streaming request  
> happened. Probably should just overload EnlargeOutputRequestedRegion  
> to set to the largest.
>
> 3) m_UpdataBuffer seems redundant with the output image.
>
> Just from my quick look.
>
> Brad
>
> ========================================================
> Bradley Lowekamp
> Lockheed Martin Contractor for
> Office of High Performance Computing and Communications
> National Library of Medicine
> blowekamp at mail.nih.gov
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.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-developers

--
Luca Antiga, PhD
Head, Medical Imaging Unit,
Biomedical Engineering Department,
Mario Negri Institute.
mail: Villa Camozzi, 24020, Ranica (BG), Italy
phone: +39 035 4535-381
email: antiga at marionegri.it
web: http://villacamozzi.marionegri.it/~luca

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20090310/bdc43249/attachment.htm>