Atmospheric Tomography Using Convolutional Neural Networks

• C. González-Gutiérrez ^{[5] [4]} ; O. Beltramo-Martin ^[1] ; J. Osborn ^[2] ; José Luis Calvo-Rolle ^[3] ; Cos Juez, F.J. de ^{[5] [4]}
  1. [1] Aix-Marseille University

     Aix-Marseille University

     Arrondissement de Marseille, Francia

     
  2. [2] Durham University

     Durham University

     Reino Unido

     
  3. [3] Universidade da Coruña

     Universidade da Coruña

     A Coruña, España

     
  4. [4] Universidad de Oviedo

     Universidad de Oviedo

     Oviedo, España

     
  5. [5] Instituto Universitario de Ciencias y Tecnologías Espaciales de Asturias (Oviedo)

  Mostrar afiliaciones +
• Localización: Intelligent Data Engineering and Automated Learning – IDEAL 2020. 21st International Conference: Guimarães, Portugal; November 4–6, 2020. Proceedings / Cesar Analide (ed. lit.), Paulo Novais (ed. lit.), David Camacho Fernández (ed. lit.), Hujun Yin (ed. lit.), Vol. 2, 2020 (Part II), ISBN 978-3-030-62365-4, págs. 561-569
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • We present an application of Convolutional Neural Networks (CNN) to atmospheric tomography that is required for compensating optical aberrations introduced by the atmospheric turbulence using dedicated tomographic Adaptive Optics (AO) systems.We compare the state of the art Minimum Mean Square Error (MMSE) reconstructor with a Multi-Layer Perceptron (MLP) and a CNN architecture and show that the CNN performs up to 15%–20% better than the MMSE and is more robust to atmospheric profile variations up to 10% compared to the MLP. Such results pave the way to implement CNN architectures to revisit atmospheric tomography for astronomical telescopes equipped with AO.