Flat optic multispectral filter array for visible hyperspectral imaging

Hyperspectral imaging (HSI) is a key technology that goes beyond the capabilities of typical RGB cameras enabling the capture of high-resolution spectral and spatial information. One of the more promising methods for HSI, termed snapshot, captures the entire hyperspectral datacube (spectral + spatial information) in a single frame. Snapshot HSI can be achieved using a large array of narrowband filters tiled on the detector similar to a Bayer filter. While intuitive, this method rejects most of the incident light and thus suffers from decreased signal-to-noise ratio (SNR) and increased latency. Recent studies have shown that multispectral filter arrays combined with reconstruction algorithms can overcome the throughput issue; however, these filters are challenging to fabricate in mass production. In this work, the project team will design a high-throughput multispectral array operating in the visible spectrum using flat optic, nanophotonic architectures that are amenable to high-rate manufacturing. Advanced optimization techniques will be used to design the filter arrays and a subset of downselected designs will be fabricated and characterized. We will also work with the customer to investigate and implement algorithms for the datacube reconstruction. By the end of Phase I, we expect to have a demonstrated concept, allowing us to rapidly develop an advanced prototype and increase the technology readiness level in Phase II and beyond.