Fig. 1. Compressive sampling based on AOTF

Fig. 2. Sparse-reconstruction Compressed Sensing System model

Fig. 3. Sampling matrix visualization（a）Sensing matrix A_c for conventional wavelength-by-wavelength point sampling；each row represents the spectral transmittance data of AOTF driven by a single frequency；（b）Four of the spectral transmittance curves（c）Sensing matrix A（50 of 202）for acousto-optic coded sampling based on compressed sensing；each row represents the spectral transmittance of the system at a given state of AOTF（d）Four of the equivalent spectral transmittance curves

Fig. 4. Comparison of the measured values obtained from the two methods

Fig. 5. Spectral reconstruction results of L1-MAGIC，l1_ls，and OMP，from 202 measurements of 512 spectral bands（600-855.5nm）.（a）Reference signal vs reconstruction signals in the wavelet transform domain.（b）Reference signal vs reconstruction signals in the spectral domain

Fig. 6. Reconstruction results of spectrum data of 75 materials in the spectral library. (a) Comparison of SAM metrics for the recovery results of the two reconstruction algorithms. (b) Comparison of GSAM metrics for the recovery results of the two reconstruction algorithms. (c) Comparison of PSNR metrics for the recovery results of the two reconstruction algorithms. (d) Spectral reflectance curves of 75 materials (0.714-2.5 μm)

Table 1. Spectral resolution data of the selected AOTF device

Table 2. PSNR of three different reconstruction algorithms

Table 3. SAM and GSAM of three different reconstruction algorithms in the spectral domain

Table 4. The average of spectral fidelity metrics