[Insight-users] RecursiveGaussianImageFilter

salah salah at gris.uni-tuebingen.de
Thu May 12 17:58:26 EDT 2005


Hello everybody,
 
I am using the piece of code below to smooth an itk image (of type unsigned short).
The input image has intensity values in the range 0-1940,
 
The output of the bcaster (bcaster should cast float back to unsigned short) is a dark
image due  to some voxel having the maximum intensity value (65535).
 
To debug thing thing, I displayed the float image resulting from m_Smoother. The image
is alright, but surprisingly the it has the intensity range (-3.867 -- 6241.296). I am wandering
how the negative values are resulted. I think they cause the wrongly cased output!!
 
Any ideas where do these negatives come from? how to properly smooth/cast??
 
many thanx,
Zein
 
 
 
//----------------------------------------------   cast to float
typedef itk::CastImageFilter< ImageType, FloatImageType > CastType; 

CastType::Pointer caster = CastType::New();

caster->SetInput(m_InputImage);

caster->Update();

RecursiveGaussianImageFilterType::Pointer x_Smoother;

x_Smoother = RecursiveGaussianImageFilterType::New();

x_Smoother->SetInput( caster->GetOutput() ); 

x_Smoother->SetDirection( 0 ); // apply along X

x_Smoother->SetOrder( RecursiveGaussianImageFilterType::ZeroOrder );

x_Smoother->SetNormalizeAcrossScale( true );

x_Smoother->SetSigma( GetSigma() );

x_Smoother->UpdateLargestPossibleRegion();

m_Smoother->SetInput( x_Smoother->GetOutput()); 

m_Smoother->SetDirection( 1 ); // apply along Y

m_Smoother->SetOrder( RecursiveGaussianImageFilterType::ZeroOrder );

m_Smoother->SetNormalizeAcrossScale( true );

m_Smoother->SetSigma( GetSigma() );

m_Smoother->UpdateLargestPossibleRegion();

//----------------------------------------------   re-cast back

typedef itk::CastImageFilter< FloatImageType, ImageType > BackCastType; 

BackCastType::Pointer bcaster = BackCastType::New();

bcaster->SetInput(m_Smoother->GetOutput() );

bcaster->Update();

m_SmoothedImage = bcaster->GetOutput();



More information about the Insight-users mailing list