Angle-Insensitive Broadband Multispectro-Polarimetric Encoding Based on Inverse Design of Mosaic Metasurfaces

Multimodal optical information detection allows precise characterization of light-matter interactions. Current multimodal optical detection systems often encounter challenges such as low spectral resolution, limited polarization sensitivity, or bulky designs that impede seamless integration. Here, a broadband multispectro-polarimetric filter (BMSPF) metasurface array enables simultaneous spectral is proposed and full-Stokes polarimetric imaging in a single snapshot. By combining Metal-Insulator-Metal (MIM) metasurfaces with a mosaic optimization design, angle-robust multispectro-polarimetric encoding under a 30° field-of-view is achieved. The encoded grayscale information is captured via a monochrome image sensor and decoded using a neural network. Operating across the visible-to- near-infrared range (400-1100 nm), the system attains an impressive spectral resolution of 6 nm. This method expands the application scenarios of multispectro-polarimetric imaging and provides a pathway for the manufacturability of metasurface devices.