An Ultra‐Compact Broadband Metalens‐Based Structured Light Analyzer

ABSTRACT High‐precision real‐time characterization of complex spatiotemporal light fields is crucial for super‐resolution imaging and high‐dimensional information processing. However, existing technologies still face challenges such as limited functionalities, restricted accuracy, and complicated measurement procedures. Here, a broadband metasurface‐based structured light analyzer is proposed and experimentally demonstrated. Leveraging a neural network‐driven inverse optimization model, the end‐to‐end inverse design of the metalens is achieved within a continuous geometric parameter space. By enabling comprehensive multi‐parameter detection within a monolithic metasurface architecture across the 450–650 nm range, this design offers an ultra‐compact form factor suitable for direct on‐chip integration with CMOS imaging sensors. Experimental results at 488, 532, and 633 nm reveal full Stokes reconstruction errors below 3%, with a polarization reconstruction accuracy comparable to that of commercial polarimeters. Furthermore, the reconstruction accuracy of the wavefront phase gradient is approximately 3%, supporting real‐time wavefront detection with an incident angle of up to 9°. Finally, the dynamic multi‐parameter capture of vector light fields featuring complex polarization and phase distributions is successfully demonstrated. This work provides an effective strategy for designing compact light‐field detection devices, holding promising potential for applications in structured light detection and joint polarization‐phase imaging.