Journal of Infrared and Millimeter Waves, Volume. 43, Issue 6, 833(2024)
An adaptive denoising of the photon point cloud based on two-level voxel
Figures & Tables(9)
Fig. 1. Schematic diagram of voxel topological relations：(a) the neighborhood between voxel vertices and vertices；(b) the neighborhood between voxel edges and edges；(c) the neighborhood between voxel faces and faces；(d) the adjacent relationship between all voxels；(e) the schematic diagram of voxel connectivity features
Fig. 2. Flowchart of the photon point cloud adaptive noise reduction algorithm based on two-level voxels
Fig. 3. Schematic diagram of study area and photon point cloud footprints：（a）study area；（b）distribution of photon point cloud footprints；（c）Raw photon point cloud data
Fig. 4. Photon point cloud denoising process: (a) night; (b) daytime
Fig. 6. Comparison of the performance on signal identification (left) and signal photons (right) of photon point cloud in daytime with strong noise (corresponding to Figure 5 (b-3)): (a) the algorithm of this paper; (b) improve DBSCAN; (c) improve OPTICS; (d) ATL08
Fig. 7. Statistical chart of the performance evaluation indicators
Table 1. Details of the experimental data
View tableView in ArticleTable 1. Details of the experimental data
Data Time Track number SNR Zenith Angle Remarks ATL03_20230630101112_01562007_006_01.h5 daytime gt1l 0.3017 22.72° b④ daytime gt2l 0.2954 22.75° b③ daytime gt3l 0.2901 22.78° c③ ATL03_20200105225356_01560607_006_01.h5 night gt1r 29.00 166.56° c③ night gt2r 26.18 166.53° c② night gt3r 23.83 166.50° c①
Table 2. The results of the proposed algorithm, improved DBSCAN, improved OPTICS and DRAGANN
View tableView in ArticleTable 2. The results of the proposed algorithm, improved DBSCAN, improved OPTICS and DRAGANN
Methods Evaluation daytime night all gt1l gt2l gt3l mean gt1r gt2r gt3r mean mean b④ b③ c③ c③ c② c① Two-Level Voxel P 0.986 1 0.972 9 0.984 7 0.981 2 0.988 3 0.982 7 0.976 3 0.982 4 0.981 8 R 0.965 5 0.981 4 0.968 9 0.971 9 0.970 9 0.977 1 0.982 4 0.976 8 0.974 4 F1 0.975 7 0.977 1 0.976 7 0.976 5 0.979 5 0.979 9 0.979 3 0.979 6 0.978 1 Improve DBSCAN P 0.949 4 0.933 1 0.942 7 0.941 7 0.963 1 0.952 7 0.966 8 0.960 9 0.951 3 R 0.931 5 0.938 0 0.930 9 0.933 5 0.947 0 0.948 3 0.944 8 0.946 7 0.940 1 F1 0.940 4 0.935 5 0.936 8 0.937 6 0.955 0 0.950 5 0.955 7 0.953 7 0.945 6 Improve OPTICS P 0.962 5 0.955 7 0.960 2 0.959 5 0.968 1 0.972 9 0.965 8 0.968 9 0.964 2 R 0.948 9 0.937 6 0.941 5 0.942 7 0.955 9 0.958 7 0.952 3 0.955 6 0.949 2 F1 0.955 7 0.946 6 0.950 8 0.951 0 0.962 0 0.965 7 0.959 0 0.962 2 0.956 6 DRAGANN P 0.980 5 0.978 3 0.985 7 0.981 5 0.978 4 0.980 1 0.973 9 0.977 5 0.979 5 R 0.913 1 0.907 4 0.899 6 0.906 7 0.932 8 0.911 1 0.928 2 0.924 0 0.915 4 F1 0.945 6 0.941 5 0.940 7 0.942 6 0.955 1 0.944 3 0.950 5 0.950 0 0.946 3
Tools
Get Citation
Copy Citation Text
Zhen-Hua WANG, Wu-Zhong YANG, Xiang-Feng LIU, Feng- Xiang WANG, Wei-Ming XU, Rong SHU. An adaptive denoising of the photon point cloud based on two-level voxel[J]. Journal of Infrared and Millimeter Waves, 2024, 43(6): 833
Paper Information
Category: Infrared Optoelectronic System and Application Technology
Received: Mar. 27, 2024
Accepted: --
Published Online: Dec. 13, 2024
The Author Email: Wei-Ming XU (xuwm@mail.sitp.ac.cn), Rong SHU (shurong@mail.sitp.ac.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20