Convolutional neural network (CNN) has great potentials in holographic reconstruction. Although excellent results can be achieved by using this technique, the number of training and label data must be the same and strict paired relationship is required. Here, we present a new end-to-end learning-based framework to reconstruct noise-free images in absence of any paired training data and prior knowledge of object real distribution. The algorithm uses the cycle consistency loss and generative adversarial network to implement unpaired training method. It is demonstrated by the experiments that high accuracy reconstruction images can be obtained by using unpaired training and label data. Moreover, the unpaired feature of the algorithm makes the system robust to displacement aberration and defocusing effect.

No abstract

We theoretically propose and experimentally verify a method to generate new polycyclic beams, namely concentric perfect Poincaré beams (CPPBs), by using an encoded annular phase mask. The proposed beams consisting of multiple polarization structured fields can be simultaneously generated in one concentric mode, which are respectively mapped by fundamental Poincaré sphere (PS), high-order Poincaré sphere (HOPS), and hybrid-order Poincaré sphere (HyPS). Moreover, the ring radius, numbers and polarization orders of the CPPBs at arbitrary positions on arbitrary PS are independently controlled. This work enriches the mode distributions of perfect vortex and introduces a new polarization degree of freedom, which has the potential to implement more information beyond the orbital angular momentum multiplexing in optical communication.

No abstract

No abstract