Acta Optica Sinica, Volume. 45, Issue 18, 1828017(2025)
Remote Sensing Image Simulation Method of Craters Based on Neural Radiance Fields (Invited)
Figures & Tables(10)
Fig. 5. Directly synthesized image (top row) and image after image harmonization network (bottom row)
Table 1. Comparison of training effectiveness of YOLOv5 algorithm
View tableTable 1. Comparison of training effectiveness of YOLOv5 algorithm
Dataset Method P R F1 mAP@0.5 CE2 Base 0.632 0.589 0.607 0.418 Mix without harmonization 0.678 0.606 0.656 0.469 Mix with harmonization 0.683 0.635 0.662 0.475 LRO NAC Base 0.724 0.359 0.537 0.246 Mix without harmonization 0.824 0.253 0.535 0.254 Mix with harmonization 0.739 0.458 0.607 0.300
Table 2. Comparison of training effectiveness of HRNet algorithm
View tableTable 2. Comparison of training effectiveness of HRNet algorithm
Dataset Method P R F1 CE2 Base 0.949 0.311 0.434 Mix without harmonization 0.841 0.379 0.481 Mix with harmonization 0.788 0.478 0.552 LRO NAC Base 0.636 0.391 0.464 Mix without harmonization 0.670 0.406 0.488 Mix with harmonization 0.638 0.421 0.494
Table 3. Comparison of training effectiveness of DeepMoon algorithm
View tableTable 3. Comparison of training effectiveness of DeepMoon algorithm
Dataset Method P R F1 CE2 Base 0.784 0.595 0.629 Mix without harmonization 0.801 0.586 0.642 Mix with harmonization 0.788 0.588 0.637 LRO NAC Base 0.626 0.518 0.549 Mix without harmonization 0.664 0.594 0.612 Mix with harmonization 0.658 0.590 0.607
Table 4. Comparison of image synthesis algorithms training effectiveness
View tableTable 4. Comparison of image synthesis algorithms training effectiveness
Algorithm P R F1 mAP@0.5 YOLOv5 Our method 0.683 0.635 0.662 0.475 Deep image blending 0.599 0.626 0.617 0.416 DoveNet 0.652 0.668 0.666 0.466 BargainNet 0.692 0.620 0.662 0.457 HRNet Our method 0.788 0.478 0.552 / Deep image blending 0.976 0.271 0.400 / DoveNet 0.895 0.252 0.376 / BargainNet 0.881 0.241 0.367 /
Tools
Get Citation
Copy Citation Text
Weichang Zhang, Xun Liu, Wei Li, Yongchao Zheng. Remote Sensing Image Simulation Method of Craters Based on Neural Radiance Fields (Invited)[J]. Acta Optica Sinica, 2025, 45(18): 1828017
Paper Information
Category: Remote Sensing and Sensors
Received: May. 30, 2025
Accepted: Jul. 30, 2025
Published Online: Sep. 15, 2025
The Author Email: Wei Li (wei_li_bj@163.com)
CSTR:32393.14.AOS251184
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20