[ITK-users] ASM and AAM in ITK

[Timothee Evain]

Hi everybody,

I'm currently trying to build teeth models to guide segmentation in CBCT data.

With the help of the ImagePCAShapeModelEstimator I was able to build an ASM model. But that's not enough and I would like to add the intensity information, thus building an appearance model.
Since there isn't build-in solution, I have two questions:

-Is anyone aware of a code project for AAM which could be used with itk?

-In the perspective of coding my own method, I've taken a look to the ImagePCAShapeModelEstimator code. The way the PCA is performed puzzle me. The principal component are computed by combining the data matrix to the eigenvectors of the inner product matrix, not by taking the eigenvector of covariance matrix, for computational efficiency. I don't see the equivalence between these methods. Is there somewhere I could get information on this, a math sheet or something like that ?

Tim