Application of improved 3D-BoNet to segmentation and 3D reconstruction of point cloud instances

doi:10.13474/j.cnki.11-2246.2024.0606

Abstract

Abstract: In order to better utilize point cloud data to reconstruct indoor 3D models, this paper proposes a 3D reconstruction method for indoor scenes based on 3D-BoNet-IAM algorithm. The method improves the instance segmentation accuracy of the point cloud data by improving the 3D-BoNet algorithm.For the problem of missing point cloud data, a method based on plane primitive merging optimization is proposed to fit the plane, and the new plane obtained from the fitting is used to reconstruct the building surface model. The improved effect of 3D-BoNet algorithm is verified on S3DIS and ScanNet V2 dataset, and it is proved through experiments that the algorithm of 3D-BoNet-IAM proposed in this paper improves the segmentation accuracy by 3.3% compared with the original algorithm; the modeling effect of this paper is compared with other modeling effects, and it is proved through comparisons that this paper’s modeling effect is more accurate. The method in this paper can improve the instance segmentation accuracy of indoor point cloud data, and at the same time obtain high-quality indoor 3D models.

Key words: point cloud data, 3D-BoNet-IAM, 3D reconstruction, instance segmentation, plane primitive

CLC Number:

P237

GUO Baoyun, YAO Yukai, LI Cailin, WANG Yue, SUN Na, LU Yihui. Application of improved 3D-BoNet to segmentation and 3D reconstruction of point cloud instances[J]. Bulletin of Surveying and Mapping, 2024, 0(6): 30-35.

References

[1] 赖国棣.基于倾斜摄影和激光点云的城市实景三维模型构建研究[J].科学技术创新, 2023(27):25-28.
[2] 晋帅,李煊鹏,杨凤,等.伪激光点云增强的道路场景三维目标检测[J].中国图象图形学报,2023,28(11): 3520-3535.
[3] 吕猛.基于BIM技术的工程管理优化路径探讨[J].科技创新与应用,2023,13(32): 189-192.
[4] 李安琪,郑艳,王博,等.基于3D点云深度学习上采样的建筑物精细化三维重建[J].现代信息科技,2023,7(21): 136-139.
[5] BIBER P,ANDREASSON H,DUCKETT T,et al.3D modeling of indoor environments by a mobile robot with a laser scanner and panoramic camera[C]//Proceedings of 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).Sendai:IEEE,2004: 3430-3435.
[6] YANG Bo,WANG Jianan,CLARK R,et al.Learning object bounding boxes for 3D instance segmentation on point clouds[EB/OL].2019-10-08[2023-10-02].http://arxiv.org/abs/1906.01140.
[7] Derpanis K G.Overview of the RANSAC algorithm[J].Image Rochester NY,2010,4(1): 2-3.
[8] GUO Haoyu,PENG Sida,LIN Haotong,et al.Neural 3D scene reconstruction with the manhattan-world assumption[C]//Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).New Orleans:IEEE,2022: 5501-5510.
[9] GAMAL A,WIBISONO A,WICAKSONO S B,et al.Automatic LiDAR building segmentation based on DGCNN and euclidean clustering[J].Journal of Big Data,2020,7(1): 102.
[10] HUANG Jin,STOTER J,PETERS R,et al.City3D: large-scale building reconstruction from airborne LiDAR point clouds[J].Remote Sensing,2022,14(9): 2254.
[11] WANG Weiyue,YU R,HUANG Qiangui,et al.SGPN: similarity group proposal network for 3D point cloud instance segmentation[C]//Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:2569-2578.
[12] YI Li,ZHAO Wang,WANG He,et al.GSPN: generative shape proposal network for 3D instance segmentation in point cloud[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Long Beach:IEEE,2019: 3942-3951.
[13] HOU Ji,DAI A,NIEßNER M.3D-SIS: 3D semantic instance segmentation of RGB-D scans[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Long Beach:IEEE,2019: 4416-4425.
[14] IKEHATA S,YANG Hang,FURUKAWA Y.Structured indoor modeling[C]//Proceedings of 2015 IEEE International Conference on Computer Vision (ICCV).Santiago:IEEE,2015: 1323-1331.
[15] MURA C,MATTAUSCH O,PAJAROLA R.Piecewise-planar reconstruction of multi-room interiors with arbitrary wall arrangements[J].Computer Graphics Forum,2016,35(7): 179-188.
[16] NAN Liangliang,WONKA P.PolyFit: polygonal surface reconstruction from point clouds[C]//Proceedings of 2017 IEEE International Conference on Computer Vision (ICCV).Venice:IEEE,2017: 2372-2380.