[Rtk-users] forward projectors - considering perspective view ?

[Simon Rit]

Hi,
I'm not sure I understand exactly what you mean but it seems to me
that yes, it should, i.e.
- the source position is set according to the description you have
given in your ThreeDCircularProjectionGeometry. Computation of the
source position is done in
ThreeDCircularProjectionGeometry::GetSourcePosition, you can check.
- idem for the pixel position of your detector. Computation is a bit
more subtle because it uses a matrix,
ThreeDCircularProjectionGeometry::GetProjectionCoordinatesToFixedSystemMatrix
The whole thing is for example done in rtk::JosephForwardProjectionImageFilter
Are you asking because you've observed something unexpected?
Simon

On Tue, Jun 23, 2015 at 1:30 PM, "Robert Calließ"
<Robert.Calliess at gmx.de> wrote:
> Hello,
> I've got a questions regarding to the forward projectors for the S-ART
> reconstruction.
> Let's say there are some xray projection images from an xray machine taken
> with a circular trjectory (cone beam)
> The perspective view for these real projection images have been taken into
> account, of course.
> The forward projector creates a virtual projection image of the volume to be
> reconstructed. When this forward projection
> takes place does the forward projector also take into account the
> perspective view of the real CT system ?
>
> Let's say we have a simple raycast forward projector.
> The geometry (focus, object, detector) are all given in real world units
> where the object's center is (0, 0, 0).
> By "real world units" I mean the physical dimensions from the CT system. For
> instance, if the distance between
> the object's center and the xray-source is 500mm, the the xray-source
> position is (0,0, -500.0).
>
> So the virtual projection image is created as follows:
> 1: setup ray from  xray-source to the current detector pixel
> 2: determine entry point and exit point of this ray with the physical
> bounding volume
> 3: sample the volume along the ray, from entry point to exit point
>
> Does these steps per se include the perspective view ? Since the geomtry
> (position of focus object detector) for the reconstruction is modeled
> to imitate the real CT system's geometry, it should ?
>
>
>
> best regards,
> Robert
>
>
>
>
>
>
> _______________________________________________
> Rtk-users mailing list
> Rtk-users at public.kitware.com
> http://public.kitware.com/mailman/listinfo/rtk-users
>