Metasurface imaging method detects corners directly in light

Researchers at City University of Hong Kong have devised an optical imaging technique that detects corners in scenes using specially engineered metasurfaces. The approach, inspired by how bees and woodpeckers focus on angular features, performs corner detection directly in the optical domain, bypassing digital processing. This could enable high-speed, low-power motion tracking applications.

The research employs a 4f imaging configuration with a metasurface positioned at the Fourier plane to modulate light in the angular direction. Silicon nitride serves as the metasurface material, with structural parameters optimized through geometric-phase modulation for efficient broadband performance across visible wavelengths.

The team validated the approach using polygon-shaped targets, confirming that corners appeared clearly highlighted while the system maintained stable operation across different visible wavelengths. Testing on polygon arrays demonstrated full detection-plane coverage with a large field of view. The researchers also derived parity rules from imaging analysis, explaining how a single metasurface can handle objects with varying amplitude, phase, and angular properties.