Statismo fit surface and how to calculate maximum variation from total training set

[Venkatesh Kaduru]

Hello,

Could someone please explain/clarify the following questions?

1. How does the Pointset to Image Registration method work? it is a bit confusing.

2. There is no information available or I did not understand how to calculate the (example) 90% or 98% shape variation from the training set. I am using 4000 mesh files in the training examples.

--

Thanks & Regards,

Venkatesh Kaduru

+49 17680890466

[Venkatesh Kaduru]

---------- Forwarded message ---------
From: Venkatesh Kaduru <kadur...@gmail.com>
Date: Mon, Aug 19, 2019 at 11:21 PM
Subject: Statismo fit surface and how to calculate maximum variation from total training set
To: scalismo <scal...@googlegroups.com>

Edit:

Hello,

Could someone please explain/clarify the following questions?

1. How does the Pointset to Image Registration method work in the statismo fit surface.cxx? it is a bit confusing.

2. There is no information available or I did not understand how to calculate the (example) 90% or 98% shape variation from the training set. I am using 4000 mesh files in the training examples.

--

Thanks & Regards,

Venkatesh Kaduru

+49 17680890466

[Marcel Luethi]

Hi Venkatesh,

You can find the explanation how PointsetToImageRegistration works in our paper "Gaussian process morphable models", in Section 4.1:

https://ieeexplore.ieee.org/document/8010438

To calculate the variance, you need to look at the eigenvalues of the covariance matrix. The sum of all eigenvalues corresponds to the total variance. To see how many components you need to cover, say 80% of the variance, simply take as many of the leading principal until the sum of their associated eigenvalues exceeds 80% of the total variance. For Statismo, this procedure is illustrated here:

https://github.com/statismo/statismo/blob/master/modules/core/include/ReducedVarianceModelBuilder.hxx

Best regards,

Marcel

--
You received this message because you are subscribed to the Google Groups "scalismo" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scalismo+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scalismo/CAPkLaHki3ktABkPpkgWEzt-P7a3D2CugS294GnCNFbk_d_5_HQ%40mail.gmail.com.