Abstract

This study presents a novel approach and proof-of-concept for measuring the deflection direction of a multimode fiber's tip through analysis of the shape and structure of speckle patterns. First, we utilize a pendulum-based apparatus to construct a comprehensive dataset for studying the association between speckle patterns and the deflection of a multimode fiber tip. Then, we train a convolutional neural network (CNN) model to learn the relationship between a fiber optic deformation parameter and the variations in the shape and structure of speckle patterns. Finally, the ability of the model in estimating the deflection direction is evaluated by using unseen images from the test set. Our experimental results show that speckle patterns provide a feasible solution for measuring the deflection direction of the multimode fiber's tip. Moreover, our results provide insights into the generalization capabilities of speckle patterns analysis for sensing the direction of deflection.

Machine learning for sensing with a multimode exposed core fiber specklegram sensor

Darcy L. Smith, Linh V. Nguyen, David J. Ottaway, Thiago D. Cabral, Eric Fujiwara, Cristiano M. B. Cordeiro, and Stephen C. Warren-Smith
Opt. Express 30(7) 10443-10455 (2022)

Spatially-resolved bending recognition based on a learning-empowered fiber specklegram sensor

Han Gao and Haifeng Hu
Opt. Express 31(5) 7671-7683 (2023)

Adaptive noise-resilient deep learning for image reconstruction in multimode fiber scattering

Mohammad Mohammadzadeh, Shima Tabakhi, and Mohammad R. Sayeh
Appl. Opt. 63(12) 3003-3014 (2024)

Deep learning the high variability and randomness inside multimode fibers

Pengfei Fan, Tianrui Zhao, and Lei Su
Opt. Express 27(15) 20241-20258 (2019)

Fiber directional position sensor based on multimode interference imaging and machine learning

Kai Sun, Zhenming Ding, and Ziyang Zhang
Appl. Opt. 59(19) 5745-5751 (2020)