Acta Optica Sinica, Volume. 44, Issue 19, 1911003(2024)
Analysis and Suppression of Hybrid Errors in Snapshot Diffractive Computational Spectral Imaging
Fig. 1. Principle of snapshot diffractive spectral imaging system. (a) Rotationally symmetric DOE structure; (b) schematic diagram of light propagation in diffractive-refractive hybrid system; (c) flowchart of spectral image reconstruction algorithm; (d) flowchart of diffractive image degradation and reconstruction
Fig. 2. DUF-based diffractive spectral image reconstruction algorithm and its sub-modules. (a) Iteration flowchart of DUF; (b) DST; (c) HSAB; (d) FFN; (e) HS-MSA
Fig. 3. Schematic diagrams of surface error. (a) Without surface error; (b) with surface error; (c) trend of diffraction efficiency of DOE with absolute value of random surface error
Fig. 4. Graphic structure location error in actual fabrication. (a) Surface of DOE under optical microscope; (b) local magnification of surface
Fig. 5. Alignment error of optical system. (a) Light propagation schematic; (b) PSF of selected spectral bands
Fig. 6. Defocusing error of optical system. (a) Light propagation diagram; (b) PSF of selected spectral bands
Fig. 7. Synthesizing haze in spectral datasets. (a) RGB visualization image of spectral data; (b) depth matrix predicted by Marigold; (c) spectral data with haze
Fig. 8. Designed DOE. (a) Surface of DOE; (b) PSF images of partial spectral bands
Fig. 9. Simulation reconstruction results. (a) Ground truths; (b) diffracted degraded images; (c) visualization of reconstructed spectral images; (d) spectral images from 400 to 700 nm
Fig. 10. Comparison of reconstruction results of 5 networks for a specific scene. (a) Ground truths; (b) EDSR; (c) MIRNet; (d) Restormer; (e) Res-Unet; (f) DUF-DST; (g) spectral curves in region ①; (h) spectral curves in region ②
Fig. 11. Diffractive degraded images with error or noise (up) and its reconstruction results (down). (a) Without error; (b) random surface error; (c) graphic structure location error; (d) alignment error; (e) defocus error; (f) sensor noise; (g) environmental haze noise; (h) ground truth
Fig. 12. Comparison of reconstruction results before and after error optimization. (a) Ground truth; (b) without considering error and without error; (c) without considering error but with error; (d) considering error but without error; (e) considering error and with error
Fig. 13. Setup of diffractive spectral imaging experiment. (a) Overall optical path; (b) fabricated DOE; (c) imaging system; (d) test object
Fig. 14. Results of indoor imaging experiments. (a) Captured images; (b) visualization of reconstructed spectral images; (c) spectral images from 400 to 700 nm
Fig. 15. Comparison of reconstruction results of 5 spectral image reconstruction networks for a specific scene. (a) EDSR; (b) MIRNet; (c) Restormer; (d) Res-Unet; (e) DUF-DST; (f) spectral curves in region ①; (g) spectral curves in region ②
Fig. 16. Results of outdoor imaging experiments. (a) Captured images; (b) visualization of reconstructed spectral images; (c) spectral images from 400 to 700 nm
Fig. 17. Comparison of reconstruction results of 5 spectral image reconstruction networks for a specific scene. (a) EDSR; (b) MIRNet; (c) Restormer; (d) Res-Unet; (e) DUF-DST; (f) spectral reflectance of green leaves of different plants[33]; (g) spectral curves in region ①
Fig. 18. Results of outdoor dehazing imaging experiments. (a) Captured images; (b) results obtained by reconstruction network without dehazing; (c) results obtained by reconstruction network with dehazing; (d) spectral images from 400 to 700 nm
Get Citation
Copy Citation Text
Xianmeng Shen, Renjin Shao, Suodong Ma, Donglin Pu, Chinhua Wang, Junxue Wang, Yue Ben, Chufeng Xue. Analysis and Suppression of Hybrid Errors in Snapshot Diffractive Computational Spectral Imaging[J]. Acta Optica Sinica, 2024, 44(19): 1911003
Category: Imaging Systems
Received: Apr. 22, 2024
Accepted: May. 20, 2024
Published Online: Oct. 12, 2024
The Author Email: Ma Suodong (masuodong@suda.edu.cn)
CSTR:32393.14.AOS240887