A general image misalignment correction method for tomography experiments

Towards full-stack deep learning-empowered data processing pipeline for synchrotron tomography experiments

“…These effects are usually coupled in the postprocessing phase, making the alignment task even more demanding. Conventional methods, like center of mass (CoM), 47 cross-correlation, 48 , 49 and iterative techniques, 50 , 51 , 52 , 53 , 54 are often applied for tomography stack optimization, but all have their limitations, mostly on accuracy and efficiency. Correcting coupled or mixed types of motion inaccuracy, drift, and vibration in a faster and more accurate fashion is demanding, and extensive research is being undertaken on DL image alignment in the STDPP.…”