Learning to see through multimode fibers

TL;DR

This paper demonstrates the use of deep neural networks to classify and reconstruct handwritten digits transmitted through multimode fibers, showing improved performance with a two-step reconstruction and classification approach.

Contribution

It introduces a neural network-based method for image classification and reconstruction through multimode fibers, extending the transmission distance up to 1 km.

Findings

Neural networks can classify digits from speckle patterns after fiber transmission.

Reconstruction before classification improves accuracy over direct classification.

Performance degrades with fiber length but can be mitigated with training strategies.

Abstract

We use Deep Neural Networks (DNNs) to classify and reconstruct a large database of handwritten digits from the intensity of the speckle patterns that result after the images propagated through multimode fibers (MMF). Images transmitted through fibers with up to 1km length were recovered. The ability of the network to recognize the input degraded with fiber length but the performance could be enhanced if the neural networks were trained to first reconstruct the image and then classify it rather than classify it directly from the speckle intensity.