Metasurfaces for Edge Detection and Spatial Differentiation in Free Space

Edge detection is an essential computer-vision tool for image processing, with applications ranging from autonomous driving to medical image analysis. Realization of such operations with ultrafast processing speed is critical in time-sensitive scenarios and has been demonstrated with metasurfaces, which are nanostructured surfaces capable of precisely tuning light at the subwavelength scale. These ultrathin devices have enabled numerous paradigms of spatial optical computing for performing direct analog differentiation and edge detection in the optical domain. This review provides an overview of the theoretical foundations, device architectures, and practical implementations of metasurface-based free-space optical edge detectors. We categorize recent advances by linking their physical architecture to the edge-detection mechanisms, where (1) local metasurfaces spatially modulate the wavefront for Fourier-domain filtering and point-spread-function engineering, whereas (2) nonlocal metasurfaces perform Fourier filtering through angularly-engineered dispersion from collective resonances. Progresses in forward and inverse design strategies based on metaheuristics and deep learning have accelerated the realization of multifunctional, fabrication-tolerant optical differentiators. The review concludes by outlining the challenges of metasurface spatial differentiators such as reconfigurability and system-level integration while discussing future opportunities in achieving adaptive, multimodal metasurface computing. These advances position metasurface-based edge detection as a potential futuristic technology for intelligent imaging and information processing.