[Insight-users] Problem with Memory considerations of Laplacian of Gaussian

[Bradley Lowekamp]

Hello,

You by any chance aren't still using ITKv 3.20? There have been numerous improvements to the LaplacianRecursiveGaussianImageFilter since then, including making many the filter run "InPlace", which must be enabled. However internally the filter still used a RealPixelType of a double for each pixel. 

I have also written a multi-scale laplacian blob detector. It currently only works to detect dark blob, but I have been meaning to make in more general and write an Insight Journal article about it. I have it as a little external module for ITKv4 on github[1]. This repository can be just cloned into ITKv4's Module/External directory. However, this filter will peak at least 3 images of double in memory.


[1] https://github.com/blowekamp/itkScaleSpace

Hope this helps,
Brad

On Apr 11, 2012, at 4:43 PM, karthik uppuluri wrote:

> Hi all,
> 
> I have montages of sizes 4GB to 16 GB to work on.Whenever I declare a 4GB montage as a  float image and apply Laplacian of Gaussian at different scales, it would end up consuming 50GB of my RAM space for each scale.(I used ITK smart pointers so the maximum consumption is 50GB). I think the laplacian of gaussian is atleast making 4 copies of my montage since it is float a 4GB image will end up consuming 16GB for each copy of it.I want to find local maxima points in my montage.I want to know if there is a better way to do that, considering the cost of using memory.
> 
> Thanks and regards
> Karthik Uppuluri
> _____________________________________
> Powered by www.kitware.com
> 
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
> 
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.php
> 
> 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-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20120411/a9825c1c/attachment.htm>