search

LLNL / LEAP

comprehensive library of 3D transmission Computed Tomography (CT) algorithms with Python API and fully integrated with PyTorch
https://leapct.readthedocs.io
MIT License
74 stars 8 forks source link

bad pixel correction feature #57

Closed hws203 closed 2 weeks ago

hws203 commented 2 weeks ago

@kylechampley Is there any bad pixel correction feature at Leap? I can see that if the detector has some dark or white bad pixel then I can see below result slice at a specific slice. I adapted a blur filter when it reads projection images, but that can not cover this issue(dark circle).

bad_pixel

kylechampley commented 2 weeks ago

Yes, there is. See the “outlierCorrection” algorithm. There is a demo script for this, see d14_outlierCorrection.py

hws203 commented 2 weeks ago

I checked both your outlierCorrection(A) and outlierCorrection_highEnergy(B). And I can see that the first one does not fully cover this issue, there is still dark circle. But the second one(B) can cover fully this issue, so I can not see any dark circle. Thanks for your quick comments.

kylechampley commented 2 weeks ago

You should be able to get better results from the standard outlier correction algorithm. Try using a bigger window size

hws203 commented 2 weeks ago

Yes it is. Windows size is critical factor. The size bigger than 5 of the standard outlier correction works well for my case. Thanks.

hws203 commented 2 weeks ago

I will close this issue.