Enhanced principle component method for fringe removal in cold atom images

Abstract: Many powerful imaging techniques for cold atoms are based on determining the optical density by comparing a beam image having passed through the atom cloud to a reference image taken under similar conditions with no atoms. In practice the beam profile typically contains interference fringes whose phase is not stable between camera exposures. To reduce the error of these fringes in the computed optical density, an algorithm based on principle component analysis (PCA) is often employed. However, PCA is general purpose and not tailored to the specific case of interference fringes. Here we demonstrate an algorithm that takes advantage of the Fourier-space structure of interference fringes to further reduce the residual fringe signatures in the optical density.