Fig. 1. Structure schematic diagram of the article

Fig. 2. Principle of spectral response computational hyperspectral imaging system

Fig. 3. (a) 195 CQD materials in the form of filters; (b) Transmission spectra for some of the CQD filters shown in Fig.(a) ^[44]

Fig. 4. (a) Physical diagram of perovskite quantum dot filters; (b) Spectral coding curves of perovskite quantum dot filters; (c) Spectral coding curve of the perovskite quantum dot filter prepared six batches^[45]

Fig. 5. Relationship between the spectral resolution and the number of QD filters at 1350 nm ^[46]

Fig. 6. (a) Schematic depicting generation of arbitrary EL spectra using an EL array; (b) Example of EL spectra; (c) relative weights of the corresponding spectra used to reconstruct the target spectrum in Fig.(d); (d) The dashed gray curve represents the desired target spectrum. The purple curve represents the experimentally measured spectrum from the implemented five-device EL array. The gold curve represents the designed spectrum from calculation^[49]

Fig. 7. (a) Experimental principle of computational spectral reconstruction based on liquid crystal devices; (b) Partial spectral coding in the experiment^[31]

Fig. 8. (a) The multi-layer structure that generates the response functions has 5 Si layers (blue slabs) with the same thickness of 15 μm. The distances varied to obtain 400 different structures; (b) Three representative response functions; (c) The recovered signal barely reproduces the original signal with a high noise level; (d) Both the thicknesses and the distances between layers are varied; (e) Three representative response functions; (f) The recovered signal agree well with the original signal^[33]

Fig. 9. (a) Spectrometer based on multiple scattering in disordered photon structures; (b) The spectral correlation function of light intensities averaged over all detection channels of a 25-mm-radius spectrometer; (c) A two-dimensional matrix composed of 25 spectral response curves used in the experiment ^[57]

Fig. 10. (a) Schematic diagram of the proposed ultra-spectral imager; (b) A comparison of the fidelity of spectral reconstruction using circular, square and freeform metasurfaces. The objects are 200 000 synthetic spectra composed of Gaussian distribution functions^[63]

Fig. 11. (a) Schematic diagram of the F-P filter structure; (b) Spectral transmission curves of F-P filters with different thicknesses^[72]

Fig. 12. (a) Random transmittances produced by the thin-film method; (b) The auto-covariance function of filter 5; (c) The cross-covariance function between filter 5 and filter 20^[36]

Fig. 13. (a) A time-division multiplexed wideband spectral structure encoding spectral imaging system by mechanically rotating switching filter; (b) The liquid crystal broadband spectral filter controls the switching of the spectral response curve by changing the voltage at both ends of the liquid crystal; (c) In a miniaturized liquid crystal spectral imaging system, the liquid crystal is located in the lens Fourier plane, and each pixel of the object is encoded by the same spectrum in a single sample^[32,47]

Fig. 14. (a) The base cell comprises 20×20 metasurface units; (b) The supercell comprises 18×22 base cells; (c) The multiplexing of the metasurface; (d) Select some rows from the matrix equation to form the corresponding micro-spectrometers^[78]

Fig. 15. (a) Design of a 6 × 6 micro-FPR array with SiO₂ as the material between the two FP mirrors; (b) Design where the micro-FPR array (a) is attached to the sensor; (c) Design where the micro-FPR array (a) is in front of a lens array^[82]