Shared-modality multi-bandpass filtering for plug-and-play multispectral depth imaging

In this work, we propose a shared-modality multi-bandpass filtering (SMF) method for plug-and-play multispectral depth imaging (MDI). Unlike traditional multispectral filter array (MFA) cameras that require precise pixel-level alignment of spectral filters, our approach leverages a low-cost planar multi-bandpass (MBP) filter, enabling flexible and customizable multispectral imaging through simple real-time linear computation. A key advantage of our method is that, compared to conventional parallel systems, which rely on cross-modal registration to align spectral and depth images, robust shared-modal registration is achieved by integrating batch-fabricated MBPs across parallel cameras with a common passband, while also significantly reducing system volume. As a proof of concept, we developed a prototype system comprising two SMF cameras, capable of real-time 7-band spectral and depth imaging. The system achieves a full width at half maximum (FWHM) of 10-15 nm in spectral imaging, 1600 × 1300 spatial resolution, and a depth accuracy of 1.75 mm at 30 fps. Its capability was further demonstrated in an edge-computing autonomous driving scenario, validating the system's practical effectiveness.