High-quality phase-only Fourier hologram generation with camera-in-the-loop

In this paper, we propose an optimization method for Fourier holograms that enables high-quality optical reconstruction of phase-only Fourier holograms. We define the amplitude of the input image for Fourier hologram calculation as an optimization plane within a camera-in-the-loop (CITL) framework to generate phase-only holograms with superior optical reconstruction quality. Unlike traditional CITL methods that optimize phase holograms, our proposed method optimizes input amplitudes that exhibit a high correlation with the original input images in the Fourier hologram. Leveraging this correlation, we introduce a neural network model for phase-only Fourier hologram, PoFNet, to infer the optimized input amplitudes from the original images, thereby addressing the time-consuming nature of the traditional CITL algorithm, which is hindered by repetitive calculations. During the training process, PoFNet employs our proposed CITL to account for non-ideal forward propagation, i.e., optical propagation. Optical experiments demonstrate that both our proposed CITL method and PoFNet effectively reduce noise in the Fourier optical path.