Abstract

We propose a novel dual-camera design to acquire 4Dhigh-speed hyperspectral (HSHS) videos with high spatialand spectral resolution. Our work has two key technicalcontributions. First, we build a dual-camera system thatsimultaneously captures a panchromatic video at a highframe rate and a hyperspectral video at a low frame rate,which jointly provide reliable projections for the underly-ing HSHS video. Second, we exploit the panchromatic video to learn an over-complete 3D dictionary to represent eachband-wise video sparsely, and a robust computational re-construction is then employed to recover the HSHS videobased on the joint videos and the self-learned dictionary. Experimental results demonstrate that, for the first time to our knowledge, the hyperspectral video frame rate reaches up to 100fps with decent quality, even when the incident light is not strong.