2023/09/27 by Liam Collard, Collard, Liam, Mohammadrahim Kazemzadeh +7
Engineering · Physics and Astronomy · #Advanced Optical Imaging Technologies #Applied Physics (physics.app-ph) #Data Analysis #FOS: Physical sciences #Optical Coherence Tomography Applications #Optics (physics.optics) #Random lasers and scattering media #Statistics and Probability (physics.data-an)
paper · pdf · doi:10.48550/arxiv.2309.15532
openalex publication_date 2023/09/27 · openalex created_date 2023/09/30 · openalex updated_date 2026/08/03
Artificial intelligence has emerged as promising tool to decode a phase image transmitted through a multimode fiber (MMF) by applying deep learning techniques. By transmitting tens of thousands of images through the MMF, deep neural networks (DNNs) are capable of learning how to decipher the seemingly random output speckle patterns and unveil the intrinsic input-output relationship. High fidelity reconstruction is obtained for datasets with a large degree of homogeneity, which underutilizes the capacity of the combined MMF-DNN system. Here, we show that holographic modulation can be employed to encode an additional layer of variance on the output speckle pattern, improving the overall transmissive capabilities of the system. Operatively we have implemented this by adding a holographic label to the original dataset and injecting the resulting phase image into the fiber facet through a Fourier transform lens. The resulting speckle pattern dataset can be clustered primarily by holographic label (rather than the image data), and can be reconstructed without loss of fidelity. As an application, we describe how colour images may be segmented into RGB components and each colour component may then be labelled by distinct hologram. A UNET architecture was then used to decode each class of speckle patterns and reconstruct the colour image without the need for temporal synchronisation between sender and receiver