[Insight-users] Time Performance of the ConfidenceConnected filter

[Flo]

Dear all:

 

I'm using itkConfidenceConnectedImageFilter to automatically extract a
tube-like shape that represents about 10% of a 512*512*256 dataset.

My program runs on a Linux Core 2 Duo machine with 4 Gb Ram mem. I need
about 7 minutes to segment my dataset.

 

Although thanks to ITK my segmentation is successful, I think the time
needed to segment my dataset is relatively high. Thus, I was wondering
whether I did something wrong or if such filter is known to have a poor
time-efficiency. Is there any alternatives or pre-processing that could
accelerate this segmentation process?

 

In addition, I'm using the signedmaurerDT and I realized that I had weird
results if I do not template my output image over a REAL type. Is it a
normal behavior?

 

Thanks for your help and comments.

 

Flo.

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20080209/b9fa7fd5/attachment.htm