Engineering Structure Recognition Method Based on Random Sample Consistency and Point Cloud Data Matching

View at publisher Summarise with VIRAT
Authors: Ling Jiao, LiYun Shi, Xian Li
Cited by: 0
References: 22

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Single architectural styles can no longer meet the visual esthetic demands of the public. Diverse architectural styles have placed high demands on engineering inspection methods, while traditional building inspection approaches lack the ability to ensure structural stability from a global perspective. Therefore, to enhance the overall perspective of engineering structures, this study proposes an innovative method for identifying engineering structures. This method combines Random Sample Consensus (RANSAC) with density-based noise application spatial clustering, using RANSAC to solve the problem of parameter setting relying on experience. Applying voxel downsampling to process structural unit point cloud data to improve processing efficiency and constructing a three-dimensional (3D) point cloud registration model, the optimal transformation between point clouds is achieved by defining and minimizing the objective function, realizing the visualization and analysis of multiplane states of engineering objects. The validation results on the 3D building model data set show that the segmentation accuracy of the model is 96.6%. In engineering structure recognition across multiple provinces, the average intersection over union is higher than 97%, and the overall accuracy exceeds 93%. The registration overlap rate of the research model in the self-built data set is 96%, and the recognition capability of engineering structures stabilizes after 900 iterations, with a root mean square error of 2.64 m, significantly outperforming common models. The results show that the proposed integrated method can efficiently and accurately handle complex building engineering inspections, globally analyze structural states, ensure diversified building safety, and provide effective solutions for the deep application of 3D technology in the construction field.

Fields of Study & Subjects

3D Surveying and Cultural Heritage 3D Shape Modeling and Analysis Infrastructure Maintenance and Monitoring

Publication details

Journal / source
Journal of Architectural Engineering
Volume
32
Issue
2
Publication date
25 March 2026
Language
EN