[Insight-users] itkImagePCAShapeModelEstimator problem
nawel h
nawel_h5 at hotmail.com
Tue Feb 6 09:10:19 EST 2007
Hello,
I have a question about the computation of the principal component in the
itkImagePCAShapeModelEstimator.
If I'm not wrong the eigenvectors are computed from A'A (instead of AA')
where A is a representation of the signed distance function subtracted with
the mean. To get the principal component of AA', the product between A and
the previous eigenvectors (ev of A'A) is done.
But in the code, the product is done with the signed distance function
(before the mean offset), right? Is it normal? Did I miss something?
Regards
Nawel
code:
for(unsigned int img_number =0; img_number < m_NumberOfTrainingImages;
img_number++ )
{
tempImageItA = m_InputImageIteratorArray[img_number];
for(unsigned int pix_number =0; pix_number < m_NumberOfPixels;
pix_number++ )
{
pix_value = tempImageItA.Get();
for( unsigned int vec_number = 0; vec_number <
m_NumberOfTrainingImages; vec_number++ )
{
m_EigenVectors[pix_number][vec_number] +=
(pix_value *
eigenVectorsOfInnerProductMatrix[img_number][vec_number]);
}
++tempImageItA;
}
}
_________________________________________________________________
Avec Windows Live OneCare éliminez tous les virus de votre PC !
http://www.windowslive.fr/liveonecare/default.asp
More information about the Insight-users
mailing list