Acta Optica Sinica, Volume. 43, Issue 9, 0930002(2023)
Broadband Spectral Reflectance Reconstruction Based on Improved Principal Component Analysis
Figures & Tables(15)
Fig. 4. Reconstructed reflectance of SG140 color card. (a) Path #16; (b) path #51; (c) path #61; (d) path #123
Fig. 5. RMSE histograms of reconstructed reflectance of SG140 color cards before and after the introduction of
Fig. 6. Paper dyed with different pigments. (a) Titanium dioxide; (b) ochre; (c) cyanine; (d) stone green
Fig. 7. Reconstructed reflectance results of dyed paper. (a) Titanium dioxide; (b) ochre; (c) cyanine; (d) stone green
Fig. 9. Reconstructed reflectance of certain points on the surface of the oil painting. (a) Reflectance of point A on the surface of the oil painting; (b) reflectance of point B on the surface of the oil painting; (c) reflectance of point C on the surface of the oil painting; (d) reflectance of point D on the surface of oil painting
Fig. 10. RMSE histograms of samples to be tested. (a) Proposed method; (b) PCA method; (c) weighted pseudo inverse method
Fig. 11. Comparison of color reconstruction effect. (a) Color reconstructed by standard reflectance; (b) color reconstructed by proposed method; (c) color reconstructed by PCA method; (d) color reconstructed by weighted pseudo inverse method
Table 1. Accuracy comparison of reflectance reconstruction of SG140 color card
View tableTable 1. Accuracy comparison of reflectance reconstruction of SG140 color card
No. Proposed method PCA Weighted pseudo inverse RMSE GFC ISSD RMSE GFC ISSD RMSE GFC ISSD 1
2
3
4
0.01299 0.99987 0.03862 0.11041 0.97804 0.28773 0.03268 0.99881 0.09058 0.05505 0.99367 0.00710 0.07794 0.99328 0.10803 0.05221 0.99758 0.07285 0.02384 0.99993 0.02486 0.03802 0.99924 0.02010 0.01354 0.99991 0.00751 0.00961 0.99943 0.01829 0.03204 0.992534 0.01510 0.02559 0.99550 0.02600
Table 2. Accuracy comparison of reflectance reconstruction of dyed paper
View tableTable 2. Accuracy comparison of reflectance reconstruction of dyed paper
No. Proposed method PCA Weighted pseudo inverse RMSE GFC ISSD RMSE GFC ISSD RMSE GFC ISSD Ⅰ 0.05747 0.99911 0.05452 0.08432 0.99690 0.06624 0.05404 0.99827 0.02686 Ⅱ 0.02165 0.99677 0.05604 0.03253 0.99413 0.12449 0.03559 0.99014 0.08449 Ⅲ 0.03279 0.98741 0.02815 0.03636 0.99036 0.15416 0.03065 0.99324 0.13417 Ⅳ 0.04312 0.99466 0.13414 0.07674 0.99758 0.26927 0.06321 0.99829 0.22072
Table 3. Accuracy comparison of reflectance reconstruction of oil painting surface
View tableTable 3. Accuracy comparison of reflectance reconstruction of oil painting surface
Point Proposed method PCA Weighted pseudo inverse RMSE GFC ISSD RMSE GFC ISSD RMSE GFC ISSD A 0.00941 0.99819 0.01159 0.01955 0.99679 0.12635 0.01099 0.99843 0.03531 B 0.03801 0.99029 0.03888 0.12056 0.96333 0.20296 0.09712 0.95924 0.02711 C 0.03350 0.99863 0.02260 0.06742 0.99669 0.06320 0.07236 0.99701 0.08552 D 0.01967 0.99660 0.01655 0.03695 0.99206 0.12976 0.01519 0.99804 0.03345
Table 4. Color difference values of CIEDE2000 using three methods
View tableTable 4. Color difference values of CIEDE2000 using three methods
Method Path #16 Path #51 Path #61 Path #123 Proposed 0.9954 4.0280 1.2679 1.2605 PCA 5.3812 6.1978 6.6691 5.3960 Weighted pseudo inverse 2.0916 2.8403 2.3527 4.1635
Tools
Get Citation
Copy Citation Text
Hai Zhao, Hongning Li, Hao Chen, Yaru Gao, Xin Yang. Broadband Spectral Reflectance Reconstruction Based on Improved Principal Component Analysis[J]. Acta Optica Sinica, 2023, 43(9): 0930002
Paper Information
Category: Spectroscopy
Received: Dec. 8, 2022
Accepted: Jan. 11, 2023
Published Online: May. 10, 2023
The Author Email: Li Hongning (lihongning_ynnu@126.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20