Abstract
Stereoscopic imaging reveals the 3D morphology of objects by collecting dense optical data from multiple views. However, traditional active methods rely on structured light illumination, while passive methods require substantial bulky optical components, hindering the development of portable, real-time stereoimaging systems. Here, we demonstrate miniaturized passive snapshot polarimetric stereoscopic imaging (SPSIM) using a polarimetric metalens. This protocol outperforms commercial polarimetric cameras in efficiency and enables the separation of circular polarization (CP) components for full Stokes parameter acquisition. By incorporating polarimetric physical priors into the neural network, we significantly enhance the accuracy of the 3D reconstruction process, enabling instant stereoimaging from a single shot with low mean absolute error. The inclusion of CP information improves 3D surface detail reconstruction, achieving depth precision within 0.15 mm. Our SPSIM system is well suited for integration into miniaturized devices for use in extreme environments, while advancing next-generation high-resolution 3D imaging systems.
Published Version
Join us for a 30 min session where you can share your feedback and ask us any queries you have