LDnet: lightweight distilled network for single-exposure digital holographic reconstruction

Digital holography technology has received widespread attention due to its ability to record and reproduce three-dimensional information of objects, particularly demonstrating enormous potential in biomedical imaging and materials science. However, traditional holographic reconstruction methods require the use of phase shift operation in the time or space domain combined with complex computational processes to achieve, which to some extent limits the application areas of holographic technology. This paper proposes a lightweight network for holographic reconstruction using knowledge distillation technology. By reducing the number of network layers and utilizing knowledge distillation to improve the performance of simple models, high-quality holographic reconstruction is achieved with only one hologram, greatly reducing the number of parameters in the network model.