Automated block segmentation of retaining walls from 3D point clouds using depth maps

Yoshiyuki Yamamoto

doi:10.51094/jxiv.3602

##article.authors##

Yoshiyuki Yamamoto 愛知工業大学

DOI:

https://doi.org/10.51094/jxiv.3602

キーワード:

retaining wall、 block segmentation、 3D point cloud、 depth map、 frequency analysis

抄録

This study establishes and validates an automated pipeline for segmenting individual retaining-wall blocks from terrestrial laser scanning (TLS) point cloud data and for scoring block-level anomalies without per-wall manual retuning. The method automatically extracts the wall region, generates a dense depth map, detects block boundaries by combining depth and RGB edge cues, and estimates block pitch and stagger pattern through autocorrelation and 2D FFT frequency analysis. The detected edge peaks are then used to construct a regular block grid and quantify anomalies at the block level. Validation on two retaining walls showed that the pipeline developed on Wall A could be applied to Wall B with the same preset hyperparameters, achieving a mean IoU of 0.928 against manual annotations (0.950 for interior patches and 0.893 for edge patches). Comparison with SAM further indicates that depth-based segmentation is more suitable than RGB-based segmentation for manufactured retaining-wall blocks with uniform texture.

利益相反に関する開示

The author declares no conflicts of interest associated with this study.

ダウンロード *前日までの集計結果を表示します

ダウンロード実績データは、公開の翌日以降に作成されます。

引用文献

Ministry of Land, Infrastructure, Transport and Tourism, Road Bureau, National Road and Technology Division: Inspection Guidelines for Road Earthwork Structures, June 2018. (in Japanese)

Riveiro, B., DeJong, M.J. and Conde, B.: Automated processing of large point clouds for structural health monitoring of masonry arch bridges, Automation in Construction, Vol.72, pp. 258–268, 2016.

Valero, E., Bosch´ e, F. and Forster, A.: Automatic segmentation of 3D point clouds of rubble masonry walls, and its application to building surveying, repair and maintenance, Automation in Construction, Vol. 96, pp. 29–39, 2018.

Galanakis, D., Maravelakis, E., Pocobelli, D.P., Vidakis, N., Petousis, M., Konstantaras, A. and Tsakoumaki, M.: SVD-based point cloud 3D stone by stone segmentation for cultural heritage structural analysis – The case of the Apollo Temple at Delphi, Journal of Cultural Heritage, Vol.61, pp. 177–187, 2023.

Rabbani, T., van den Heuvel, F.A. and Vosselman, G.: Segmentation of point clouds using smoothness constraint, International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 36, Part 5, pp. 248–253, 2006.

Schnabel, R., Wahl, R. and Klein, R.: Eﬃcient RANSAC for point-cloud shape detection, Computer Graphics Forum, Vol. 26, No. 2, pp. 214–226, 2007.

Qi, C.R., Su, H., Mo, K. and Guibas, L.J.: PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 652–660, 2017.

Smith, J. and Paraskevopoulou, C.: 3D Deep Learning for Segmentation of Masonry Tunnel Joints, Procedia Structural Integrity, Vol. 64, pp.220–227, 2024.

Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.-Y., Doll´ ar, P. and Girshick, R.: Segment Anything, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp. 4015–4026, 2023.

Itakura, K., Hayashi, T. and Takata, Y.: Segmentation of stonewalls from 3D models for stonewall BIM creation, Artificial Intelligence and Data Science, Vol. 6, No. 3, pp.632–640, 2025. (in Japanese)

Matsushita, K., Gokyu, K., Yamamoto, Y., Nakamura, G. and Nakamura, E.: Estimation of track center based on arrangement characteristics of railway space using terrestrial laser scanner, Journal of Japan Society of Civil Engineers, Vol. 81, No. 22, 24-22012, 2025. (in Japanese)