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
83 stars 9 forks source link

Remove the limitation of 5 degree aligned to z-axis #72

Closed hws203 closed 1 week ago

hws203 commented 1 week ago

@kylechampley, I did check your d32_laminography.py at your latest v1.16. version And I can see good and fast result even I use the leapct.FBP() function for the reconstruction at the 5 degree tilt of detector's row without iterative method. So I tried to remove some returning code at leapct to see the result from more over 5 degree of FBP(). But there is some error in ramp filter processing of leapct. Could you help me how to remove this 5 degree limitation for my test? Because the recon time is very critical for my case.

kylechampley commented 1 week ago

Hello. I don't want to remove that restriction for everyone, but if you want to remove it yourself, it is actually pretty easy. Just change line 1071 of src/parameters.cpp.

OK with you? Let me know how it goes.

hws203 commented 1 week ago

Oh it is easy, now I can see my test result at over 5 degree. Thanks for your comments. It is ok for me.

kylechampley commented 1 week ago

Great! How does the reconstruction look?

hws203 commented 1 week ago

Your doFBP also shows a good result after laminography reconstruction with 50 degree angle as like below. If the geometry params are correct, then fbp also makes a good slice too.

oct_coil
hws203 commented 1 week ago

Below is your demo laminography result with fbp and 10 degree angle.

lamino_50_degree