SASSI — Super-Pixelated Adaptive Spatio-Spectral Imaging

We introduce a novel video-rate hyperspectral imager with high spatial, temporal and spectral resolutions. Our key hypothesis is that profiles of pixels within each super-pixel tend to be similar. Hence, scene-adaptive spatial sampling scene, guided by its segmented image, capable obtaining high-quality reconstructions. To achieve this, we acquire an RGB image the compute super-pixels, from which generate mask locations where measure high-resolution spectrum. The subsequently estimated fusing measurements using learnable filtering approach. Due low computational complexity superpixel estimation step, our setup can capture images scenes little overhead over traditional snapshot cameras, but significantly higher validate proposed technique extensive simulations as well lab prototype measures video at resolution $600 \times 900$ 600 × 900 pixels, 10 nm visible wavebands, achieving frame rate 18fps.