[Insight-users] Supervised classification with ITK

[Luis Ibanez]

Hi Franz,


Thanks for your detailed question.


Let me give it a shot at populating the table that you proposed:


Classification Algorithms in ITK
=======================

> - Supervised SVM (Support Vector Machines)
     (No, unless you count NN as a potential implementation,
      in which case, please see C2 below)

> - Unsupervised k-Means clustering (Yes) (See C1)

> - Supervised SOM classification
     (Self-Organizing Maps)
     (Yes)(See C2)

> - Supervised Linear Discriminant Classifier (Is it?)
     (Not sure on this one, it looks like it could be
      implemented by using the mixture model classifier...
      See C3)

> - Supervised Maximum Likelihood (only founded Unsupervised)
     (Not sure on this one. it would seem that you could
      implement it by using the classifier framework,
      and initializing the classes, See C4).

> - Supervised Minimum distance or K-means (No)

> - K-NN (available in the Insight Journal, See C5)

> - Parzen Classifier (No)


------------------------------------------------------------------

C1:
===
http://public.kitware.com/Insight/Doxygen/html/classitk_1_1ImageKmeansModelEstimator.html
http://public.kitware.com/Insight/Doxygen/html/classitk_1_1ScalarImageKmeansImageFilter.html
Insight/Code/Review/Statistics
itkKdTreeBasedKmeansEstimator.h
itkKdTreeBasedKmeansEstimator.txx

C2:
===
Neural Networks
Insight/Code/Numerics/NeuralNetworks/


C3:
===
http://public.kitware.com/Insight/Doxygen/html/classitk_1_1Statistics_1_1ExpectationMaximizationMixtureModelEstimator.html
http://public.kitware.com/Insight/Doxygen/html/classitk_1_1Statistics_1_1MixtureModelComponentBase.html


C4:
===
Insight/Testing/Code/Review/Statistics/itkImageClassifierFilterTest.cxx


C5:
===
http://www.insight-journal.org/browse/publication/94
"Computational Geometry Computation and KNN Segmentation in ITK"
by Cardenes R., Sanchez M.R., Ruiz-Alzola J.


-----------------


That said, please note there is a separate toolkit
called the

                      Orfeo Toolbox
           http://www.orfeo-toolbox.org/

That is build on top of ITK and is targeted to remote
sensing applications.

OTB provides support for:

http://www.orfeo-toolbox.org/SoftwareGuide/SoftwareGuidech17.html#x44-24800017


 17.3 Support Vector Machines
http://www.orfeo-toolbox.org/SoftwareGuide/SoftwareGuidech17.html#x44-26100017.3

 17.4 Kohonen’s Self Organizing Map
http://www.orfeo-toolbox.org/SoftwareGuide/SoftwareGuidech17.html#x44-27100017.4


You may find interesting the following thread in their forum
http://groups.google.com/groups/profile?enc_user=jXRgfxcAAADBXgn_3JhBl33kZTl3A-5YHqZiDvCVswhrZ6TQxKj0ww




   Please let us know if you have further questions,



         Regards,


                  Luis


-----------------------------------------------------------------------------------
On Thu, May 13, 2010 at 6:57 AM, Franz Schiller
<franzschiller1975 at gmail.com> wrote:
> Greetings
> My name is Franz Schiller and I'm doing a LandCover Classification project
> at my host-Research Centre (Goddard Space Flight Centre). My objective is to
> compare Supervised classification algorithms for a certain landcover. I
> found that ITK has a huge library of ssegmentation,filters and
> classification algorithms so I'm willing to use it. I just need to know if
> the following methods are included in ITK, and for that I'm requesting your
> help to indicate me if this supervised classification algorithms are
> available at ITK:
> - Supervised SVM (YES)
> - Unsupervised k-Means clustering (YES)
> - Supervised SOM classification (YES)
> - Supervised Linear Discriminant Classifier (Is it?)
> - Supervised Maximum Likelihood (only founded Unsupervised)
> - Supervised Minimum distance or K-means (is it?)
> - K-NN (is it?)
> - Parzen Classifier (didn't find anything)
> Could anyone give give me an indication of yes/no and in case yes, could you
> send me a class name or something that can be useful for me, in order to
> start working with it? My idea is to use a single application/library to
> compare different methods...
> Thank you for your help
> Best regards,
> Franz Schiller
> Univ Maryland
>
> _____________________________________
> 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.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-users
>
>