[TuP-C] Poster Session 1

[TuP-C-9] Numerical Investigation of Bending Tolerance of Multimode Optical Fiber Imaging with Variational Autoencoder

○Hana Fowler¹, Masatoshi Bunsen², Taiki Otani¹ (1. Graduate School of Engineering, Fukuoka University (Japan), 2. Department of Electronics Engineering and Computer Science, Fukuoka University (Japan))

Keywords:

Other novel fibers,fiber devices and applications,Emerging non-telecom applications including sensing and imaging,Artificial intelligence and machine learning for optical transmission systems and subsystems

We propose a Variational Autoencoder-based image reconstruction through a multimode optical fiber. Simulations under bending conditions achieved 96% accuracy, demonstrating the feasibility even with fiber deformation. This contributes to robust multimode optical fiber imaging.