[Insight-users] How to reduce the needed RAM for distance map creation?

[lynx.abraxas at freenet.de]

On 10/08/09 17:19:02, lynx.abraxas at freenet.de wrote:
> Anyway, with the 4.4GB dataset I get the following:
> 
> time ~/itk/watershed07/watershed07 zs_gauss-0.5-9_ru_16bit.raw zs_gauss-0.5-9_ru_ws_113101.tif 1 1 3 1 0 1 1550 1549 950
> signed danielson distance map...
> Morphological watershed...
> Creating output...
> terminate called after throwing an instance of 'itk::ExceptionObject'
>   what():  /home/ttd/localusr/include/InsightToolkit/Common/itkImportImageContainer.txx:188:
> Failed to allocate memory for image.

Since  I  can now use 8-bit encoding for my binarised data and cropping I'm at
1.2GB of data. I noticed by  inserting  Update()  that  the  creation  of  the
distance map is consuming a lot of memory (around 50 times as much as the size
of the input dataset). I assume that depends on the actual length of  occuring
distances in the evaluation. It seems that after a slow steady increas to 12GB
RAM usage there was a sudden  need  of  so  much  more  RAM  that  my  program
terminated within the update period of top. (And so I never saw it use up more
than 26% of RAM.)
Is there any way I can reduce the needed memory further? I can't crop more nor
would  I  like  to  bin  the  data. I know that commercial programs can create
distance maps (perhaps less accurate) with much less memory consumption.
Is there any way I can estimate the maximum memory needed and also the time to
finish computing?

Can  I  free the memory the distance map filter needed when it has finished so
there is more RAM again for the following filters?

Regards,
Lynx