Presentation + Paper
13 December 2020 Convolutional neural networks for object-agnostic wavefront sensing in the presence of noise
Author Affiliations +
Abstract
This paper shows that a simple convolutional neural network (CNN) can be used to build an object-agnostic wavefront sensor. Using the well-known Phase Diversity approach as a point of departure, Fourier-space metrics are computed from the conventional and diversity images and then fed to the CNN, which predicts values of the underlying Zernike coefficients. The methodology is shown to work in the presence of Gaussian noise. Prediction errors for defocus, astigmatism, and spherical are on the order of 1/100 of the wavelength.
Conference Presentation
© (2020) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Kedar R. Naik, Raymond H. Wright, David D. Claveau, D. Scott Acton, and J. Scott Knight "Convolutional neural networks for object-agnostic wavefront sensing in the presence of noise", Proc. SPIE 11448, Adaptive Optics Systems VII, 114481H (13 December 2020); https://doi.org/10.1117/12.2576147
Lens.org Logo
CITATIONS
Cited by 1 scholarly publication.
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Wavefront sensors

Convolutional neural networks

Network architectures

Adaptive optics

Image processing

Monochromatic aberrations

Spherical lenses

Back to Top