Fig. 1: Conceptual sketch of metasurface-empowered hyperspectral imaging. | Nature Communications

Fig. 1: Conceptual sketch of metasurface-empowered hyperspectral imaging.

From: Metasurface-empowered snapshot hyperspectral imaging with convex/deep (CODE) small-data learning theory

Fig. 1

The developed off-axis focusing meta-mirror, composed of multi-resonant meta-atoms, can simultaneously image a color object and spatially separate the image into four wavelength channels in free space. The acquired 4-band multispectral image will be used to construct an 18-band hyperspectral data cube using the CODE small-data learning and imaging theory inspired by the convex optimization (CO) in conjunction with the deep learning (DE) technique.

Back to article page