By Simon Göppel

Inverse problems, such as medical imaging and non-destructive testing, require stabilizing the reconstruction process to account for measurement errors. We propose a new method, called translation invariant diagonal frame decomposition (TI-DFD), that overcomes image distortions caused by classical wavelet frames. By combining TI-DFD with a regularizing filter, we achieve accurate and stable solutions with optimal convergence rates. Our approach eliminates wavelet artifacts and shows promise for various applications.

For more information: https://iopscience.iop.org/article/10.1088/1361-6420/accd15