[ITK-users] [ITK] Vectorize/Linearize itk::Image

[Timothee Evain]

Hi Sarthak,

That's not really a generic framework so I doubt you will find a ready-to-use filter to do that.
If you want to hide all the processing code you may want to write your own filter.
On a side note, if you're only interested in intensities as output, and willing to discard the correspondance between position and value, why not just compute the otsu threshold, then parse the buffer, adding the value to your vector when it's over/under this threshold (depending on what you consider as background)?
That would be less hassle than a full mask image.

HTH
Tim


----- Mail original -----
De: "Scapegoat Sarthak" <scapegoat.sarthak at gmail.com>
À: "Bradley Lowekamp (NIH/NLM/LHC) [C]" <blowekamp at mail.nih.gov>, dzenanz at gmail.com
Cc: insight-users at itk.org
Envoyé: Mardi 24 Mai 2016 16:05:28
Objet: Re: [ITK] [ITK-users] Vectorize/Linearize itk::Image

Hi Bradley, Dzenan, 

Thanks for your replies. The main idea is the following: 

1. Read image from file -- inputImage 
2. Generate a mask to remove background information (either using Otsu or reading another image with this information) -- inputMask 
3. Vectorize the inputImage in the region defined by inputMask -- tempColMat or tempRowMat 
4. Concatenate multiple tempColMat/tempRowMat -- trainingData 
5. Pass trainingData to classifier of choice. 

I have tried both the solutions presented and neither work as optimally as I would like. Ideally, I would like to use a single multi-threaded filter which takes a mask as input along with the mask and extracts pixel values as a vector in return. 

Best, 
Sarthak 


On 24 May 2016 at 09:55, Lowekamp, Bradley (NIH/NLM/LHC) [C] < blowekamp at mail.nih.gov > wrote: 


Hello, 

You want to look at the ImportImageContainer[1]. Every itk::Image has one of these object to hold the buffer. The fastest way to “convert” the image, is not to do any work and just get a pointer to the buffer. 

HTH, 
Brad 

[1] https://itk.org/Doxygen/html/classitk_1_1ImportImageContainer.html 

> On May 24, 2016, at 9:48 AM, Scapegoat Sarthak < scapegoat.sarthak at gmail.com > wrote: 
> 
> Hello All, 
> 
> What would be the fastest way to convert an itk::Image< DataType, Dimensions > to a single vector of intensities? Ideally, I would like to concatenate multiple such vectors and pass them to machine learning algorithms. 
> 
> Thanks, 
> Sarthak 
> _____________________________________ 
> 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://public.kitware.com/mailman/listinfo/insight-users 



_____________________________________
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://public.kitware.com/mailman/listinfo/insight-users

_______________________________________________
Community mailing list
Community at itk.org
http://public.kitware.com/mailman/listinfo/community