Robust and automated computational adaptive optics using multiple randomized sub-apertures

Abstract

Computational adaptive optics (CAO) is emerging as an attractive alternative to hardware-based solutions for diffraction-limited optical coherence tomography, e.g., of the human retina. Still, to become a reliable and robust solution, many challenges need to be solved. Here, we present CAO based on multiple randomized sub-apertures in combination with suitable filtering to remove disturbing artifacts. We show that this approach can reliably detect aberrations, and we compare results to other algorithms, such as optimization of imaging quality. We also demonstrate that the filtering of reflecting image structures is essential for a robust determination of aberrations.