Deep learning-based video stream reconstruction in mass-production diffractive optical systems

Authors:
  • SUSamara National Research University
    Samara National Research University
  • VEV. Evdokimova
    V. Evdokimova
  • MPM. Petrov
    M. Petrov
  • MKM. Klyueva
    M. Klyueva
  • IRIpsi Ras - Branch Of The Fsrc “Crystallography And Photonics” Ras
    Ipsi Ras - Branch Of The Fsrc “Crystallography And Photonics” Ras
  • EZE. Zybin
    E. Zybin
  • FSFederal State Unitary Enterprise State Research Institute of Aviation Systems
    Federal State Unitary Enterprise State Research Institute of Aviation Systems
  • VKV. Kosianchuk
    V. Kosianchuk
  • IMI. Mishchenko
    I. Mishchenko
  • VNV. Novikov
    V. Novikov
  • NSN. Selvesiuk
    N. Selvesiuk
  • EEE. Ershov
    E. Ershov

Abstract
Many recent studies have focused on developing image reconstruction algorithms in optical systems based on flat optics. These studies demonstrate the feasibility of applying a combination of flat optics and the reconstruction algorithms in real vision systems. However, additional causes of quality loss have been encountered in the development of such systems. This study investigates the influence on the reconstructed image quality of such factors as limitations of mass production technology for diffractive optics, lossy video stream compression artifacts, and specificities of a neural network approach to image reconstruction. The paper offers an end-to-end deep learning-based image reconstruction framework to compensate for the additional factors of quality losing. It provides the image reconstruction quality sufficient for applied vision systems.
Funding Information
  • Russian Science Foundation (20-69-47110)
  • Russian Foundation for Basic Research (18-07-01390-А)
  • Ministry of Science and Higher Education of the Russian Federation (007-ГЗ/Ч3363/2б)

This publication has 24 references indexed in Scilit: