Low cross-talk multiple-image hiding via lateral-shifting multiplexing

Abstract

We propose a lateral-shifting multiplexing method for multi-image hiding with low cross-talk. Unlike schemes that use a single degree of freedom (such as wavelength $\lambda$, distance z, or angle $\theta$), our approach translates phase keys across two degrees of freedom ($\Delta x$ and $\Delta y$), enabling the successful extraction of high-quality images with a 1 mm multi-plane spacing. Meanwhile, the integration of a physics-driven neural network enhances the system's ability to reveal hidden images. Compared to diffraction distance multiplexing scheme with a single degree of freedom (z), our approach reduces cross-talk between extracted images, as demonstrated through experimental validation using a cascaded phase system.