Overview and Choice of Artificial Intelligence Approaches for Night-Time Adaptive Optics Reconstruction

Adaptive optics (AO) is one of the most relevant systems for ground-based telescopes image correction. AO characterized by demanding computational that must be able to quickly manage large amounts data, trying make all calculations needed closest real-time. Furthermore, next generations are already being constructed will demand higher requirements. For these reasons, artificial neural networks (ANNs) have recently become alternative commonly used tomographic reconstructions based on several algorithms as least-squares method. ANNs shown its capacity model complex physical systems, well predicting values in case nocturnal where some models been tested. In this research, a comparison terms quality outputs given and time presented between three common ANN topologies nowadays, obtain fits better Multi-layer perceptron (MLP), convolutional (CNN) fully (FCN) considered. The results determine way forward development reconstruction future telescopes, ones under construction solar observations.