Lensless complex amplitude demodulation based on deep learning in holographic data storage

To increase the storage capacity in holographic data (HDS), information to be stored is encoded into a complex amplitude. Fast and accurate retrieval of amplitude phase from reconstructed beam necessary during readout HDS. In this study, we proposed demodulation method based on deep learning single-shot diffraction intensity image verified it by non-interferometric lensless experiment demodulating four-level phase. By analyzing correlation between features encoding pages, inverse problem was decomposed two backward operators denoted convolutional neural networks (CNNs) demodulate respectively. The experimental system simple, stable, robust, only needs single realize direct both our investigation, first time HDS that multilevel achieved experimentally one without iterations.