[MINC-users] minctracc - Order of Transformation using -lsq12

[Andrew Janke]

Hi Elizabeth,

I too had this same question a while back as I was trying to reverse
engineer these parameters in perl for rotations larger than 90deg.
90deg is the max that minctracc will handle before it will give up.

I just put this code on github here:

   https://github.com/BIC-MNI/minc-widgets/blob/master/xfmdecomp/xfmdecomp.pl

In answer to your question, the bet would be to look at the code:

   https://github.com/BIC-MNI/mni_autoreg/blob/master/minctracc/Numerical/make_rots.c#L236

You will note a comment in the header which gives you the answer you want:

   mat = (T)(C)(S)(SH)(R)(-C)


a


On 30 May 2014 02:14, Elizabeth Wicks <elizabethw at msmri.medicine.ubc.ca> wrote:
> Hi all,
>
> I am wondering the exact order of the components in the /minctracc -lsq12/
> transformation. That is, in what order are the rotation, translation,
> scaling and shear components done during the registration? For example, is
> the output matrix = rotation_matrix*skew_matrix*scale_matrix? And what are
> the forms of these individual matrices, e.g. is rotation_matrix = Rx*Ry*Rz
> or Rz*Ry*Rx?
>
> In particular, it would be helpful to know how /minctracc/ creates its
> output xfm matrix, given the individual transformation parameters,
> (essentially the breakdown of what /param2xfm/ does).
>
> Thanks in advance!
>
> Elizabeth
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users