[ITK] Cost Function Evaluations at MultiResolutions

[wassim bouzeid]

Hi Everyone,

I want to determine the rotational transformations corresponding to the extrema of the cost function at the first level of resolution (both images are subsampled by a factor of 8 and interpolated). 

I am using an exhaustive optimizer (itk::ExhaustiveOptimizerv4) to explore the parametric space using VersorTranfrom (only 3D rotations). 

The parametric space is explored using large rotation angles (coarse grid) with large step size . For example, If I have three values per dimension, there is 125 possible configurations  for my parameter settings ((-20,-20,-10),(-20,-20,-10)...(-20,-20,10)). 

//*******************************Here is my code*********
typedef itk::ExhaustiveOptimizerv4< double > ExOptimizerType;
ExOptimizerType::StepsType stepsLevel8( initialTransformCOM->GetParameters() );
  stepsLevel8[0] = 3;
  stepsLevel8[1] = 3;
  stepsLevel8[2] = 3;
  optimizerLevel8->SetNumberOfSteps( stepsLevel8 );
  optimizerLevel8->SetStepLength( 10 );
//*******************************Here is my code*********

I am feeling that I am missing something such the range that the optimizer has to explore. Is this the correct way to do it?

Any help would be appreciated.

Thanks,
Sam
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/community/attachments/20160615/a70ae4c7/attachment.html>