Open murdock-grewar opened 10 months ago
I had a quick look at the fan beam projection code.
I believe this issue may have something to do with the switch_t
and associates, because the discontinuities are at 3*PIdiv4
, 5*Pidiv4
, 7*PIdiv4
, 9*PIdiv4
and so on.
I can reproduce the issue, but not quite sure about the origin yet. line_fanflat
projector seems to work fine.
Hi. There appears to be a bug in astra's fan-beam projection code. I have included image examples and source code to produce a particularly potent example.
Examples
First, here is the volume being projected (it is nonnegative everywhere):
Here is a sinogram I produced with since-lost settings. Each row corresponds to a different projection angle. There are discontinuities between rows. Here is another sinogram, where the issue is much more prevalent. There are even negative values, as can be seen from the colour scale on the right-hand side.
Source code
Here is a
.py
file which reproduces the second sinogram: