Hi,
Steve is right - the 7th DOF is a global scaling.
I also wanted to add that quaternions do not specify
what kind of transformation you are applying - they are
a way of representing angles, much in the same way as
Euler angles are. Hence you need to know more about
what they mean in ITK by a "Quaternian3DTransform".
If you hadn't said, I would have assumed that this would
be a rigid-body transformation as it is a standard
convention to use quaternions to represent rotations
alone, and so this plus translations would give a 6 dof
rigid-body transformation. My guess is that any 7 dof
transformation (using quaternions or not) is probably
the same as ours: rigid-body + global scaling. But you
should double check this with any available ITK information.
All the best,
Mark
On 3 Dec 2004, at 19:33, Martin Kavec wrote:
> Thanks for the link, Christian.
>
> The whole lecture is really great. However, I did't find quite what I
> expected.
> I'll try to put what I am looking for in another words.
>
> I work on a quite difficult registration problem and I use FLIRT
> results as a
> refence. My implementation platform is ITK ( http://www.itk.org ) and
> my aim is
> to replicate, at the begining, FLIRT results. ITK implements three
> types of
> transformation for 3D registration: AffineTransform (DOF=12),
> Quaternian3DTransform (DOF=7) and Versor3DRigidTransform (DOF=6).
>
> So I was wondering what transforms FLIRT uses.
>
> Cheers,
>
> Martin
>
> Quoting "Christian F. Beckmann" <[log in to unmask]>:
>
>> Martin,
>>
>> the FSL course talk on registration etc covers this, see
>> http://www.fmrib.ox.ac.uk/fslcourse/lectures/flirt_fugue/s_1030xfm.htm
>>
>> cheers
>> christian
>>
>>
>>
>> On 3 Dec 2004, at 14:13, Martin Kavec wrote:
>>
>>> Hi.
>>>
>>> I am trying to figure out, what transform is optimized with
>> different
>>> DOFs in 3D registration. In DOF=12, I would guess, it is Affine
>>> transform.
>>> I am particularly interested in DOF=6, 7. I briefly looked at the
>>> FLIRT source code, but it was probably too brief. Is it a Quaternion
>> in
>>> DOF=7?
>>>
>>> Thanks for explanation.
>>>
>>> Martin
>>>
>>>
>> --
>> Christian F. Beckmann
>> Oxford University Centre for Functional
>> Magnetic Resonance Imaging of the Brain,
>> John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
>> Email: [log in to unmask] -
>> http://www.fmrib.ox.ac.uk/~beckmann/
>> Phone: +44(0)1865 222782 Fax: +44(0)1865 222717
>>
|