[Rtk-users] ThreeDCircularProjectionGeometry Angular Arc

[Simon Rit]

Hi,
You're right, it's better to cover slightly more than 360 degrees if the
scanner follows the same piece of trajectory at the beginning and at the
end (on a circle since you're using FDK). Rotation angles are currently
forced between 0 and 2 pi, see here
<https://github.com/SimonRit/RTK/blob/master/src/rtkThreeDCircularProjectionGeometry.cxx#L103>.
I think you can feed RTK with unordered projections, it will order them to
compute angular gaps between the projections. See code here
<https://github.com/SimonRit/RTK/blob/master/src/rtkThreeDCircularProjectionGeometry.cxx#L430-L484>.
My advice is to quickly write a simulation code with a sample geometry to
check that you obtain the expected results.
Keep us posted if something does not work as expected,
Simon

On Wed, Dec 9, 2020 at 7:19 PM Benjamin W. Maloney via Rtk-users <
rtk-users at public.kitware.com> wrote:

> Hi all,
>
> I had a question about the angular arc of projections fed into
> the ThreeDCircularProjectionGeometry to be used in an FDK Cone Beam
> Reconstruction.
>
> The hardware for the CT system I'm using is not capable of moving exactly
> 360 degrees as part of a scan. If I know the angular step well is it better
> for the projections to cover slightly less or slightly more than 360
> degrees?
>
> Can RTK handle arcs larger than 1 full rotation or arcs slightly less than
> 1 full rotation without errors?
>
> My assumption is that slightly largely is best to avoid artifacts but I
> want to verify that RTK can handle larger rotations without errors
>
> Best,
> Ben
>
> _______________________________________________
> Rtk-users mailing list
> Rtk-users at public.kitware.com
> https://public.kitware.com/mailman/listinfo/rtk-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://public.kitware.com/pipermail/rtk-users/attachments/20201209/23c62c95/attachment.htm>