Abstract
A radial carpet (RC) optical beam is a type of structured light beam that is classified as a subfamily of combined half-integer Bessel-like beams. Coherent detection of such structured modes can be difficult given their structural complexity and the deterioration of the quality of modes during propagation in turbulent environments. In this paper, we first study the detection of RC modes by using fork-shaped diffraction gratings and then propose an incoherent detection approach to classify 16 classes of RC modes based on training a convolutional neural network model using DenseNet-201 architecture. The dataset comprises recorded images of RC modes after a 120 cm propagation length in a turbulent underwater environment. As the evaluations indicate, the classifier has an accuracy of 98% in identifying RC modes measured in moderate turbulence with a strength level of ${ C}_{ n}^{2}\sim{{10}^{- 12}}\;{{\rm m}^{- 2/3}}$. This method eliminates the difficulties caused by the use of multiple optical elements in coherent detection techniques such as diffraction gratings. Besides simplifying the optical system settings, it also reduces the volume and cost of implementation, especially in optical communication applications.
© 2023 Optica Publishing Group
Full Article | PDF ArticleMore Like This
Mahdi Khodadadi Karahroudi, Mohammad Khodadadi Karahroudi, Abolhasan Mobashery, and Bahman Parmoon
Appl. Opt. 58(8) 1886-1894 (2019)
Saifollah Rasouli and Somaye Fathollazade
J. Opt. Soc. Am. B 41(3) 728-737 (2024)
Saifollah Rasouli, SAEED HAMZELOUI, and Davud Hebri
Opt. Express 27(13) 17435-17448 (2019)