Abstract: This code corrects fringe artefacts in near-infrared astronomical images taken with old generation CCD cameras. It essentially solves a robust PCA problem, masking out astrophysical sources, and models the contaminants as a linear superposition of (unknown) modes, with (unknown) projection coefficients. The problem uses nuclear norm regularization, which acts as a convex proxy for rank minimization.
The code is written in python, using cupy for GPU acceleration, but will also work on CPUs.
Credit: Prunet, Simon; Bu, An-Yi