This paper proposes a deep-learning approach for Jones matrix imaging using photon arrival data, overcoming the limitations of traditional coincidence measurement techniques under low-light conditions. A variational autoencoder (β-VAE) is trained to extract minimal data representation, subsequently transformed into a Jones matrix image. The method achieves high accuracy with significantly fewer photons than previous semi-analytic algorithms, automating algorithm formulation and assessing experimental information sufficiency.
Publisher
npj Nanophotonics
Published On
Mar 06, 2024
Authors
Jiawei Xi, Tsz Kit Yung, Hong Liang, Tan Li, Wing Yim Tam, Jensen Li
Tags
deep learning
Jones matrix
imaging
photon arrival data
variational autoencoder
Related Publications
Explore these studies to deepen your understanding of the subject.
