Correcting optical wavefront distortion due to strong atmospheric turbulence using deep neural networks

In both observation astronomy and (quantum) optical communication, one of the methods widely used to correct optical wavefront distortion due to atmospheric turbulence is using phase holograms. This task relies heavily on the ability to decompose the distorted wavefront to find the coefficients of Zernike polynomials, which is resource-intensive and time-consuming. Here, we propose a novel convolutional neural network model to efficiently and accurately perform Zernike decomposition of optical wavefront distortion up to 12 Zernike modes. We demonstrate that this deep learning-based method can provide an efficient computational framework for real-time correction of optical communication in the presence of strong atmospheric turbulence.