Spectral routers for snapshot multispectral imaging

Abstract

Snapshot spectral imaging aims at capturing full spatial and spectral information simultaneously. It can benefit greatly from a photon-efficient spectral decomposition that does not sacrifice spatial resolution. The multispectral filter arrays that enable compact single-chip snapshot multispectral imaging, however, suffer from fundamental and important limitations: low photon efficiency and low spatial resolution. Here, we introduce a spectral router for snapshot multispectral imaging to address these limitations. A spectral router routes all light incident on its entire surface directly to the photodetector of each spectral channel without need for a propagation layer between the router and the detector. Unlike filters, spectral routers do not reject light to achieve spectral selectivity and can ideally exploit 100% of the incident light. Spectral routers also break the size barrier that exists for filter arrays and can be designed with wavelength size footprint. This enables spectral routers to simultaneously provide spectral information as well as spatial information at higher spatial resolution, and can even allow them to provide spectral information without sacrificing spatial information beyond the diffraction limit. We illustrate the concept with two examples. A first spectral router covers the visible range and can achieve ideal optical efficiency for six spectral channels in a sub-micrometer footprint, which is 15 times smaller than a filter array repeat unit with the smallest Fabry–Pérot filters. The second example covers the shortwave infrared (SWIR) range with nine spectral channels and is suitable for pixels that are two times smaller than the smallest SWIR imager pixels demonstrated to date.