CERN / TIGRE

TIGRE: Tomographic Iterative GPU-based Reconstruction Toolbox
BSD 3-Clause "New" or "Revised" License
560 stars 183 forks source link

FDK weights minor bug #335

Closed AnderBiguri closed 2 years ago

AnderBiguri commented 2 years ago

A user reported that these weigths:

https://github.com/CERN/TIGRE/blob/1571e3296c2351163b281b5f242440f6ddcf2465/Common/CUDA/voxel_backprojection.cu#L244-L245

should be instead (note the minus):

realx=-(geo.sVoxelX-geo.dVoxelX)*0.5f  +indX*geo.dVoxelX   +xyzOffset.x; 
realy=-(geo.sVoxelY-geo.dVoxelY)*0.5f  +indY*geo.dVoxelY   +xyzOffset.y+COR; 

And I think that is correct. Re-check, but fix if needed.