Fig. 1. Preprocessing flow chart of 3D point clouds for neural network
Fig. 2. Schematic diagram of voxel grid algorithm
Fig. 3. Subject point cloud and noise point cloud
Fig. 4. Model architecture for point clouds classification
Fig. 5. Delaunay triangulation. (a) Insert point; (b) judgement of the common inner edge using empty circumcircle; (c) delete the inner edge; (d) insert completed
Fig. 6. Overall flow chart
Fig. 7. Example diagrams of 3D models. (a) Sphere; (b) cube; (c) cylinder; (d) penetrant; (e) open-source table point cloud set; (f) open-source chair point cloud set
Fig. 8. Comparison of denoising effect of the standard ball. (a) k-neighbor denoising algorithm; (b) radius filtering algorithm; (c) proposed denoising algorithm
Fig. 9. Comparison of denoising effect of the standard gauge block: (a) k-neighbor denoising algorithm; (b) radius filtering algorithm; (c) proposed denoising algorithm
Fig. 10. Comparison of denoising effect of the standard cylinder: (a) k-neighbor denoising algorithm; (b) radius filtering algorithm; (c) proposed denoising algorithm
Fig. 11. Comparison of denoising effect of the open-source table point cloud set: (a) k-neighbor denoising algorithm; (b) radius filtering algorithm; (c) proposed denoising algorithm
Fig. 12. Processing flow
Fig. 13. Relationship between reconstruction number and reconstruction time when the number of point clouds is 100000
Fig. 14. Relationship between reconstruction number and reconstruction time when the number of point clouds is 700000
Fig. 15. Relationship between reconstruction number and reconstruction time when the number of point clouds is 1200000
Point cloud set | k-neighbor denoising algorithm | Radius filtering algorithm | Proposed denoising algorithm |
---|
Standard ball | 103642 | 108392 | 88349 | Standard gauge block | 671400 | 687593 | 659876 | Standard cylinder | 1125163 | 1149893 | 1116856 | Open-source table point cloud set | 451306 | 452099 | 429705 |
|
Table 1. Number of point clouds after denoising by the three methods
Number of initial point clouds | Traditional preprocessing method | Proposed preprocessing method |
---|
Denoising time ta1/s | Simplifying time ta2/s | Total time ta/s | Simplifying time tb1/s | Denoising time tb2/s | Total time tb/s |
---|
108957 | 0.973 | 0.025 | 0.998 | 0.053 | 0.026 | 0.079 | 728907 | 1.880 | 0.067 | 1.947 | 0.094 | 0.134 | 0.228 | 1263084 | 9.460 | 0.238 | 9.698 | 0.306 | 0.589 | 0.895 |
|
Table 2. Comparison of the time required for point cloud preprocessing
Number of initial point clouds | Traditional preprocessing method | Proposed preprocessing method |
---|
Number of point clouds after denoising | Number of point clouds after simplifying | Number of point clouds after simplifying | Number of point clouds after denoising |
---|
108957 | 103642 | 1063 | 1521 | 1346 | 728907 | 671400 | 4789 | 6113 | 5897 | 1263084 | 1125163 | 12040 | 20963 | 20738 |
|
Table 3. Comparison of point cloud number after preprocessing
Test No. | Test accuracy a1 | Test accuracy b1 |
---|
1 | 0.904 | 0.907 | 2 | 0.901 | 0.905 | 3 | 0.913 | 0.921 | 4 | 0.909 | 0.904 | 5 | 0.917 | 0.905 |
|
Table 4. Comparison of accuracy
Number of 3Dpoint clouds | Time required by the traditional method t/s | Time required by the proposed method t'/s |
---|
10 | 10.08 | 0.87 | 20 | 21.78 | 2.18 | 40 | 41.54 | 3.92 | 60 | 62.12 | 5.98 | 80 | 82.46 | 7.64 | 100 | 101.94 | 9.54 | 120 | 122.08 | 11.98 |
|
Table 5. Comparison of reconstruction time when the number of point clouds is 100000
Number of 3D point clouds | Time required by the traditional method t/s | Time required by the proposed method t'/s |
---|
10 | 20.33 | 2.36 | 20 | 42.08 | 4.78 | 40 | 82.74 | 9.52 | 60 | 123.56 | 14.34 | 80 | 164.14 | 19.26 | 100 | 204.06 | 23.78 | 120 | 245.28 | 29.06 |
|
Table 6. Comparison of reconstruction time when the number of point clouds is 700000
Number of 3Dpoint clouds | Time required by thetraditional method t/s | Time required by the proposed method t'/s |
---|
10 | 97.97 | 9.03 | 20 | 194.59 | 17.94 | 40 | 393.72 | 37.06 | 60 | 588.54 | 56.49 | 80 | 780.97 | 71.96 | 100 | 980.56 | 92.44 | 120 | 1178.41 | 109.98 |
|
Table 7. Comparison of reconstruction time when the number of point clouds is 1200000