[ITK-users] Silly question on itk::Matrix

[vincent ngai]

Hi Jim, thanks for the background, that clarified things for me.

Regards,
Vincent


On Sat, Aug 23, 2014 at 8:12 PM, Jim Miller <millerjv at gmail.com> wrote:

> Matrices and images are row major order.  Points and vectors are column
> vectors.
>
> No transpose is necessary.  You can index element by element directly to
> set the column you want.
>
> Row major ordering is more popular than column major ordering. I think
> there are regional preferences as to the"proper" way. I also think
> electrical engineers prefer one way whereas mechanical engineers prefer the
> other. Probably is partially based on how disciplines view the composition
> of coordinate frames (inside out or outside in).
>
> Some systems allow you to use either convention. We found that leads to a
> lot of confusion and errors. So 15 years ago, when we started ITK, we chose
> to pick a single standard for the code.
>
> Jim
>
> On Aug 22, 2014, at 9:49 PM, vincent ngai <vincent.ngai at gmail.com> wrote:
>
> Ah yes, so my question is why is the [] accessor not column major? If
> points are columns, the matrix elements for the Tx,Ty,Tz translation
> components should also be columnar? It just seems like extra work to need
> to get the matrix transpose just to do a Mat * P
>
> Regards,
> Vincent
>
>
> On Fri, Aug 22, 2014 at 7:14 PM, Jim Miller <millerjv at gmail.com> wrote:
>
>> ITK standard is that points are column vectors. Thus, matrix times a
>> point a point is defined as Mat * P
>>
>> Jim
>>
>> > On Aug 22, 2014, at 3:30 AM, vincent ngai <vincent.ngai at gmail.com>
>> wrote:
>> >
>> > Hi all, pardon me for this silly question on itk::Matrix. Why do I need
>> to transpose a translation matrix after I copy the [x,y,z] translation
>> values to the matrix[3].
>> >
>> > I know conceptually it should not really matter whether its row or col
>> major so long as you can
>> > post multiply or pre multiply with another vector. But the way
>> operator*() works only allows for
>> > Matrix * P, and when I test the simple case of TranslationMatrix *
>> Point, I would have to transpose the matrix. If I could also do Point *
>> TranslationMatrix, then i wouldn't need to transpose.
>> >
>> > Any ideas?
>> >
>> > --
>> > Regards,
>> > Vincent
>> > _____________________________________
>> > Powered by www.kitware.com
>> >
>> > Visit other Kitware open-source projects at
>> > http://www.kitware.com/opensource/opensource.html
>> >
>> > Kitware offers ITK Training Courses, for more information visit:
>> > http://www.kitware.com/products/protraining.php
>> >
>> > Please keep messages on-topic and check the ITK FAQ at:
>> > http://www.itk.org/Wiki/ITK_FAQ
>> >
>> > Follow this link to subscribe/unsubscribe:
>> > http://public.kitware.com/mailman/listinfo/insight-users
>>
>
>
>
> --
> Regards,
> Vincent
>
>


-- 
Regards,
Vincent
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/insight-users/attachments/20140825/423b1480/attachment.html>