Author Affiliations
Faculty of Geomatics, East China University of Technology, Nanchang , Jiangxi 330013, Chinashow less
Fig. 1. Flow chart of point cloud filtering
Fig. 2. Diagram of verticality
Fig. 3. Diagram of height difference constraint and distance condition constraint
Fig. 4. Diagram of connected graph. (a) Connected graphs without constraints; (b) connected graphs with constraints
Fig. 5. Two dimensional grid diagram of seed points
Fig. 6. Add new ground seed points to the blank grid
Fig. 7. Diagram of distance from point to fitted plane
Fig. 8. Point cloud data filtering results. (a) DSM of raw data; (b) true DEM; (c) DEM of the filtering results of the proposed method; (d) error distribution of filtering results of the proposed method
Fig. 9. Comparison of total errors of the five methods
Fig. 10. Comparison of mean values of the three kinds of error of the five methods
Fig. 11. Average total errors of different verticality thresholds
Fig. 12. Average total errors of different height difference thresholds
Fig. 13. Average total errors of different distance thresholds
Environment | Site | Sample | Feature |
---|
City | 1 | 11 | Hillsides, low vegetation, buildings | 12 | Hillsides, buildings | 2 | 21 | Large buildings, bridges | 22 | Irregular structure | 23 | Large irregular structure | 24 | Steep sides | 3 | 31 | Complex building complex | 4 | 41 | Blank data | 42 | Train tracks and trains | Country | 5 | 51 | Steep sides, low vegetation, blank data | 52 | 53 | 54 | 6 | 61 | Roads, buildings, data gaps | 7 | 71 | Bridges, roads, an underground passage |
|
Table 1. 15 groups of point cloud data and their characteristics
Category | Filtering result |
---|
Number of ground points | Number of object points |
---|
Reference result | Number of ground points | | | Number of object points | | |
|
Table 2. Error matrix
Sample | /% | /% | /% | |
---|
Average | 5.72 | 9.29 | 5.44 | 0.81 | Sample 11 | 26.74 | 13.26 | 21.04 | 0.58 | Sample 12 | 7.92 | 1.39 | 4.76 | 0.90 | Sample 21 | 2.71 | 2.94 | 2.76 | 0.92 | Sample 22 | 3.21 | 15.60 | 7.03 | 0.83 | Sample 23 | 8.00 | 4.33 | 6.27 | 0.87 | Sample 24 | 6.10 | 10.06 | 7.15 | 0.82 | Sample 31 | 0.37 | 1.85 | 1.05 | 0.98 | Sample 41 | 1.01 | 4.63 | 2.75 | 0.94 | Sample 42 | 11.10 | 0.43 | 3.55 | 0.91 | Sample 51 | 1.33 | 6.73 | 2.42 | 0.92 | Sample 52 | 3.09 | 17.35 | 4.41 | 0.75 | Sample 53 | 6.36 | 19.34 | 6.84 | 0.44 | Sample 54 | 2.46 | 3.91 | 3.24 | 0.94 | Sample 61 | 3.47 | 7.51 | 3.57 | 0.56 | Sample 71 | 1.93 | 30.03 | 4.81 | 0.72 |
|
Table 3. Filtering error of 15 experiment data of the proposed method