END-TO-END DESIGN OF LOW-COST COMPUTATIONAL TELESCOPES

Abstract

Computational imaging has transformed mobile photography, enabling high-performance imaging (e.g. HDR, Portrait mode, Night Sight) in a compact form factor that fits in our pockets. These advances are powered by the core strategy in computational imaging - the co-optimization of optical hardware and post-processing software. For example, our DiffuserCam lensless camera is a compact and inexpensive 3D imager that consists only of a scattering element (a diffuser) placed on top of an image sensor, combined with postprocessing algorithms. Here, we will explore the feasibility of using similar concepts to push the limits of three fundamental imaging problems of relevance to the Air Force, with applications in both ground and space-based telescopes. First, we will focus on the problem of digital aberration correction. Current state-of-the-art deconvolution techniques are limited by the loss of information due to aberration-induced blurring and are onerous to implement for shift-varying systems. However, by introducing new image formation models for rotationally-symmetric systems and co-designing algorithms and hardware modifications (e.g. inserting a diffuser), we will show that it is possible to circumvent the information loss to achieve higher resolution in practice. Second, we will use these ideas in a lensless camera design to demonstrate large-aperture imaging with a tiled array of commercial sensors placed behind a diffuser. Such a setup could lay the foundation for novel fold-out space telescope designs that use compact optics and inexpensive hardware, then correct the images computationally, for improved cost to performance ratio. Finally, we will design new lensless strategies for compact high-resolution hyperspectral imaging that can circumvent tradeoffs between spatial and spectral resolution via compressed sensing. Throughout this project, our designs will be optimized in an end-to-end fashion, using machine learning tools to solve for the optimized diffuser surface profiles.

Document Details

Document Type
DoD Grant Award
Publication Date
Apr 20, 2023
Source ID
FA95502210521

Entities

People

  • Laura Waller

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of California Regents

Tags

Fields of Study

  • Physics

Readers

  • Computer Vision.
  • Optical Physics and Photonics.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space
  • Space - Space Objects