Acta Optica Sinica, Volume. 43, Issue 10, 1030001(2023)
A Method for Natural Spectral Reproduction Based on Fully Connected Neural Network
Fig. 1. Structure of fully connected neural network
Fig. 2. Training and spectral reproduction in the fully connected network. (a) Process of spectral reproduction; (b) change curve of correlation index
Fig. 3. Comparison of standard solar spectra of AM1.5, CIE-D65, and CIE-A and reproduction spectra. (a) AM1.5; (b) CIE-D65;(c) CIE-A
Fig. 4. Comparison of measured natural spectra and reproduction spectra at different time on the same sunny day. (a) 6:00; (b) 11:00; (c) 18:00; (d) 19:30
Fig. 5. Comparison of spectrum reproduction between fully connected neural network model and genetic algorithm. (a) Coincidence comparison; (b) fitting time comparison
|
|
Get Citation
Copy Citation Text
Zimao Ren, Huimin Lu, Liya Feng, Lu Yang, Yifan Zhu, Jianping Wang. A Method for Natural Spectral Reproduction Based on Fully Connected Neural Network[J]. Acta Optica Sinica, 2023, 43(10): 1030001
Category: Spectroscopy
Received: Aug. 31, 2022
Accepted: Jan. 29, 2023
Published Online: May. 9, 2023
The Author Email: Lu Huimin (hmlu@ustb.edu.cn)