Single-shot reconstruction of high-dimensional complex orbital angular momentum spectrum under system misalignment

The detection of orbital angular momentum (OAM) is fundamental for advancing various applications involving vortex beams. Current measurement methodologies face challenges such as inefficiencies, complex system configurations, and the requirement for high interference stability. The notable challenges of OAM reconstruction are system misalignment and single-shot measurement. We propose a model-driven artificial neural network approach to reconstruct high-dimensional complex OAM spectra from single-shot diffraction intensity measurement under system misalignment, eliminating the need for extensive experimental data for training. Our approach may advance OAM-based high-dimensional information encoding and optical communication while also fostering further exploration at the intersection of physical models and neural networks.