| Re: Radon parameters [message #40215] |
Tue, 27 July 2004 00:30  |
Timm Weitkamp
Messages: 66
Registered: August 2002
|
Member |
|
|
On 26.07.04 at 18:28 +0100, Tim Yates wrote:
> Dear all,
>
> This sort of question has come along before, but I'm still not able to give
> the correct parameters to the RADON function.
>
> The situation is as follows. I have 73 experimentally obtained projections
> of an object, each 193 by 193 pixels. This should give me 193 sinograms,
> each 193 by 73 pixels. For each sinogram, I wish to produce a
> backprojection, and I can do this using the obsolete RIEMANN command, but
> I'd like to do so with the RADON function eg,
>
> IDL> bp = RADON(/backprojection, sinogram, THETA=theta, RHO=rho)
>
> The 73 projections were taken over an angular range -70 degrees to +74
> degrees, with 2 degree increments. I am thus setting 'theta' using,
>
> IDL> theta=(2*FINDGEN(73) - 70) * !DTOR
>
> But, what of rho? I think that this should be a vector of 73 elements, but I
> can't fathom the rationale behind how it should be formed. RADON reports
> that "ARRAY, RHO, or THETA dimensions are incompatible", and I'm pretty sure
> the RHO's to blame.
Tim,
rho should be a named vector with as many elements as there are pixels in
each (1D) projection. Thus, if the sinograms are centered around the axis
of rotation, then you should set rho to (FINDGEN(193) - 192 / 2.0)
(otherwise you may apply an offset to this that allows for the
eccentricity).
The number of elements of rho can be one of the reasons for the error
message you receive; another might be the dimension ordering of the
sinogram. Sinograms passed to RADON should have the dimensions [<no of
pixels>, <no of projections>]. Thus, if the error persists, try
backprojecting TRANPOSE(sinogram) instead of sinogram.
Hope this helps,
Timm
--
Timm Weitkamp
Paul Scherrer Institut, 5232 Villigen PSI, Switzerland
|
|
|
|
| Re: Radon parameters [message #40297 is a reply to message #40215] |
Wed, 28 July 2004 02:40  |
Tim[1]
Messages: 4
Registered: June 2004
|
Junior Member |
|
|
Dear Timm,
Thanks for the info. Transposing the sinogram fixed the problem.
One may use RADON or RIEMANN for backprojection, but which do you find that
there are particular advantages in using one over the other? It strikes me,
for one thing, that my routine with RIEMANN runs faster than the equivalent
with RADON.
Best,
Tim
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
