Chirality-driven all-optical image differentiation.

Optical analog computing enables powerful functionalities, including spatial differentiation, image processing, and ultrafast linear operations. Yet, most existing approaches rely on resonant or periodic structures, whose performance is strongly wavelength-dependent, imposing bandwidth limitations and demanding stringent fabrication tolerances. Here, to address some of these challenges, we introduce a highly tunable platform for optical processing, composed of two cascaded uniform slabs exhibiting both circular and linear birefringence, whose response exhibits features relevant to optical processing without relying on resonances. Specifically, using a coupled-wave theory framework we show that sharp reflection minima, referred to as spectral holes, emerge from destructive interference between counter-propagating circularly polarized waves in uniform birefringent slabs, and can be engineered solely through parameter tuning without requiring any spatial periodicity. When operated in the negative-refraction regime enabled by giant chirality, the interference response acquires a highly parabolic form around the reflection minimum, giving rise to a polarization-selective Laplacian-like operator that performs accurate spatial differentiation over a broad spatial-frequency range. This functionality is demonstrated through an edge-detection proof of concept. The required material parameters align closely with recent experimental demonstrations of giant, tunable chirality via meta-optics, presenting a promising pathway towards compact and reconfigurable platforms for all-optical pattern recognition and image restoration.