Reconfigurable metasurface enabled by ferroelectric superdomain for AI-driven shape completion.

Area of Science:

• Optics and Photonics
• Materials Science
• Artificial Intelligence

Background:

• Metasurfaces enable optical system miniaturization and advanced functionalities.
• Challenges exist in extracting optical information and reconstructing geometry from obscured objects.

Purpose of the Study:

• To propose a novel infrared metasurface for all-optical computing and 3D shape reconstruction.
• To demonstrate tunable optical properties and robust geometric modeling from incomplete data.

Main Methods:

• Integration of continuous graphene with a periodic ferroelectric domain array for tunable transmittance.
• Reconfiguration of ferroelectric domains to tune spectral response (8-17 μm).
• Application of a shape completion network with encoded optical transmission for 3D reconstruction.

Main Results:

• Achieved tunable relative transmittance from negative to positive values (-6% to 10%) under low gate voltages.
• Demonstrated robust 3D shape completion from sparse and incomplete point clouds using passive thermal imaging.
• Recovered occluded structural details to achieve high-resolution geometric models.

Conclusions:

• The developed reconfigurable metasurface merges advanced optical computing with AI-driven reconstruction.
• This approach offers potential for intelligent vision systems in autonomous navigation and 3D sensing.