Neural Achromatic Imaging for Large‐Scale Metalens

ABSTRACT Metalenses leverage subwavelength nanostructures to achieve high‐resolution imaging through phase manipulation, offering a flat alternative to traditional refractive and diffractive imaging systems. However, diffraction‐induced chromatic aberration remains a longstanding obstacle to full‐color, large‐scale metalens imaging. In this work, we propose a neural achromatic imaging framework that integrates a meta–achromatic network with a physics‐informed, PSF‐aware mechanism. The meta–achromatic network discovers and exploits cross‐channel image priors through PSF‐aware attention to enable high‐quality achromatic imaging. Experimental results demonstrate a 51% reduction in an introduced image‐based chromatic aberration metric and a 10.19 dB improvement in peak signal‐to‐noise ratio for imaging with a 7‐mm‐aperture metalens. The proposed framework can potentially be extended to other imaging schemes, spectral regimes, and diffractive computational imaging systems, thereby correcting chromatic aberrations and expanding the practical scale of flat optics.