Table 3 Comparison of RANSAC and RANSAC-LP4 performance on the Open3D dataset. The table reports the mean percentage improvement in the number of inliers detected relative to the baseline (RANSAC-109), the mean rank of each algorithm, and the mean percentage of the original point cloud covered by inliers. Higher mean improvements and better ranks indicate superior performance. Algorithms shown in italics represent versions of our method.

From: Optimizing plane detection in point clouds through line sampling

Algorithm

Mean improvement

Mean rank

Mean point cloud coverage

RANSAC-LP4-957

12.01% ± 10.05

1.33 ± 0.51

35.96% ± 12.15

RANSAC-LP4-747

11.69% ± 9.91

2.01 ± 0.74

35.87% ± 12.13

RANSAC-LP4-558

11.30% ± 9.66

2.84 ± 0.75

35.76% ± 12.15

RANSAC-LP4-388

10.42% ± 9.50

4.25 ± 1.18

35.50% ± 12.09

RANSAC-957

9.16% ± 9.15

6.53 ± 1.36

35.08% ± 11.86

RANSAC-LP4-239

9.13% ± 9.02

6.44 ± 2.00

35.12% ± 12.05

RANSAC-747

9.00% ± 8.97

6.77 ± 1.17

35.04% ± 11.89

RANSAC-558

8.49% ± 8.37

7.54 ± 0.99

34.94% ± 11.97

RANSAC-388

7.60% ± 7.52

8.35 ± 1.11

34.75% ± 12.11

RANSAC-239

5.69% ± 5.87

9.79 ± 1.32

34.32% ± 12.37

RANSAC-LP4-109

5.16% ± 7.81

10.64 ± 1.35

34.00% ± 11.99

RANSAC-109

0.00% ± 0.00

11.53 ± 1.21

32.94% ± 12.98