2016/06/01 by Xiaoxu Leng, Jun Xiao, Ying Wang · 3 citations
Computer Science · Earth and Planetary Sciences · Engineering · #3D Surveying and Cultural Heritage #Image and Object Detection Techniques #Robotics and Sensor-Based Localization
paper · doi:10.1111/phor.12145
crossref issued 2016/06/01 · crossref published 2016/06/01 · crossref published-print 2016/06/01 · openalex publication_date 2016/06/01 · crossref published-online 2016/06/17 · crossref created 2016/06/17 · crossref deposited 2025/06/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/25 · crossref indexed 2026/07/25
Abstract The detection of planes from 3D point clouds plays an important role in fields such as 3D modelling, rock mechanics, registration and robotics. A multi‐scale plane‐detection method is proposed, based on extracting growth units using the Hough transform and subsequent region growing into actual plane boundaries. Because the approximate geometric features of the original plane area can be obtained from the growth units, discriminant conditions for adaptive plane growing could be provided for the growing stage. In order to detect physical planes at different scales, a strategy of multi‐scale normal estimation is introduced. Simulated icosahedron point clouds, together with actual engineering point clouds, were used to test the performance of the proposed method. Experimental results show that planar point clouds could be detected effectively, especially with rock‐mass engineering applications.