[Insight-users] Problems with itkImageClassifier and itkGaussianClassifier

[Jisung Kim]

Hi.

I might be wrong. However, the following is my
understanding about the situation:

It seems that the old GaussianSupervisedClassifier has
been replaced by the
Algorithms/itkImageGaussianModelEstimator(paramter
estimation for Gaussian classes),
Algorithms/itkImageClassifierBase(integration point),
Common/itkMinimumDecisionRule (class label selection
rule), and
Code/Numerics/Statistics/itkMahalanobisDistanceMembershipFunction.
(distance function)

As you already know there is no more single
GaussianSupervisedClassifer. If you want to use the
new statistical classification framework, you have to
create proper estimator, selection rule, and
membership-function first, and then plug them into an
itkImageClassifierBase instance. After that, you can
use the instance of itkImageClassifierBase just like
using the old stand alone classifiers. There is an
test program in   Testing/Code/Algorithm,
"itkSuperVisedImageClassifierTest" that shows how to
use the framework.

If you want to use old stand alone classifiers such as
GaussianSupervisedClassifier, I believe you should
also use old version of Gibbs prior segmentation
codes.

I'm not 100% sure. So if I am wrong, please correct
me. 

Thank you,

Jisung Kim
--- Samuel_Rodr�guez_Bescos
<srodrigu@gbt.tfo.upm.es> wrote:
> Hello every body,
> I'm developing a single example based in Gibbs prior
> segmentation for grey scale images.
> I follow the steps of Dimtris Metaxas and Ting Chen
> but for grey scale image and  I have the next error:
> 
> (149) : error C2664: 'SetClassifier' : cannot
> convert parameter 1 from 'class
> itk::SmartPointer<class
> itk::GaussianSupervisedClassifier<c
> lass itk::Image<class itk::Vector<unsigned
> short,1>,2>,class itk::Image<unsigned short,2> > >'
> to 'class itk::SmartPointer<class
> itk::ImageClassifierBase<class itk::Image<class
> itk::Vector<unsigned short,1>,2>,class
> itk::Image<unsigned short,2> > >'
>         No constructor could take the source type,
> or constructor overload resolution was ambiguous
> Error executing cl.exe.
> 
> I'm using one of the last version of Insight in
> order to use the itkScalarToArrayCastImageFilter. So
> I imported the classes related with the
> itkGaussianSupervisedClassifierm from previous
> version. Has anybody any idea about what is
> happening?
> 
> If I can't find the problem. Could anybody tell me
> what class I can use to first estimation to set in
> the GibbsPriorFilter instead of gaussian Supervised
> Classifier?.
> 
> Thanks for reading to the end.
> 
> Samuel
> 
> 
> 


=====
Jisung Kim
bahrahm@yahoo.com
106 Mason Farm Rd.
129 Radiology Research Lab., CB# 7515
Univ. of North Carolina at Chapel Hill
Chapel Hill, NC 27599-7515

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com