Author Affiliations
1School of Science, Kunming University of Science and Technology, Kunming 650500, China2Yunnan Provincial Key Laboratory of Modern Information Optics, Kunming University of Science and Technology, Kunming 650500, Chinashow less
Fig. 1. Point cloud coarse denoising results
Fig. 2. Point cloud fine removal noise process
Fig. 3. Schematic diagram of slope angle calculation of point cloud data and data segmentation processing diagram,(a) the angle of the elliptical domain under different slope angles, (b) slope angle calculation, (c) data segmentation, (d) consolidation of data segments
Fig. 4. The density cluster threshold is determined
Fig. 5. Specific clustering removal noise process
Fig. 6. Point cloud data fine removal noise
Fig. 7. ICESat-2 satellite transit area in the Area of Yellowstone National Park and Great Smoky Mountains Forest Park in the United States
Fig. 8. The spaceborne photon counting lidar matches the NEON data
Fig. 9. ATLAS data matches onboard data
Fig. 10. Denoising results of different denoising algorithms: (a) (d) (g) (j) the result of the algorithm of this paper processing Data1-4, (b) (e) (h) (k) the result of the LOF algorithm processing Data1-4, (c) (f) (i) (l) the result of the DBSCAN algorithm processing Data1-4
Fig. 11. Comparison of ground curves for different data
Fig. 12. Comparison of crown curves in different data
Fig. 13. The algorithm in this paper is compared with ATL08 data,(a)、(c)、(e)、(g) the results of processing and classifying Data1-4 for the algorithm herein, (b)、(d)、(f)、(h) the ATL08 data corresponding to Data1-4
| 本文算法 | LOF算法 | DBSCAN算法 |
---|
| P | R | F | P | R | F | P | R | F | Data1 | 0.878 | 0.991 | 0.931 | 0.667 | 1 | 0.8 | 0.799 | 0.999 | 0.888 | Data2 | 0.891 | 0.999 | 0.942 | 0.848 | 1 | 0.918 | 0.877 | 1 | 0.934 | Data3 | 0.851 | 0.999 | 0.919 | 0.845 | 1 | 0.916 | 0.86 | 0.952 | 0.904 | Data4 | 0.706 | 0.966 | 0.816 | 0.429 | 0.95 | 0.591 | 0.638 | 0.974 | 0.771 |
|
Table 1. Comparison of different algorithms
| 坡度角自适应的椭圆域聚类算法 | 椭圆域聚类算法 |
---|
| P | R | F | T(s) | P | R | F | T(s) | Data1 | 0.878 | 0.991 | 0.931 | 2.531 | 0.817 | 1 | 0.899 | 16.766 | Data2 | 0.891 | 0.999 | 0.942 | 4.781 | 0.873 | 0.999 | 0.932 | 115.234 | Data3 | 0.851 | 0.999 | 0.919 | 3.453 | 0.846 | 1 | 0.917 | 52.578 | Data4 | 0.706 | 0.966 | 0.816 | 4.188 | 0.685 | 0.98 | 0.806 | 66.297 |
|
Table 2. Comparison of the efficiency of elliptic area clustering algorithms and slope adaptive elliptic area clustering algorithms
| | 本文算法 | ATL08数据 |
---|
| | (m) | | (m) | (m) | | (m) | Data1 | 地面 | 0.358 8 | 0.999 7 | -0.019 9 | 0.732 3 | 0.998 7 | 0.207 5 | | 冠层 | 3.744 9 | 0.968 6 | -1.688 7 | 17.761 | 0.276 1 | -16.850 4 | Data2 | 地面 | 0.918 | 0.999 | -0.035 2 | 1.312 | 0.998 | 0.079 7 | | 冠层 | 4.349 1 | 0.977 7 | -2.593 6 | 7.259 6 | 0.936 4 | -5.033 | Data3 | 地面 | 1.732 3 | 0.993 | -0.365 1 | 2.833 1 | 0.976 2 | 0.912 4 | | 冠层 | 4.397 4 | 0.95 | -0.868 6 | 5.110 4 | 0.94 | -0.161 4 | Data4 | 地面 | 2.177 5 | 0.999 4 | -0.568 7 | 12.121 9 | 0.980 2 | 7.887 9 | | 冠层 | 5.967 8 | 0.994 7 | -2.415 2 | 9.627 4 | 0.986 | 0.355 5 |
|
Table 3. In this paper, the algorithmic processing results are compared with the ATL08 data