Registration of airborne LiDAR point clouds by matching the linear plane features of building roof facets

• Verfasst von: Wu, Hangbin [VerfasserIn]
• Verfasst von: Fan, Hongchao [VerfasserIn]
• Titel: Registration of airborne LiDAR point clouds by matching the linear plane features of building roof facets
• Verf.angabe: Hangbin Wu and Hongchao Fan
• E-Jahr: 2016
• Jahr: 25 May 2016
• Umfang: 17 S.
• Teil: volume:8
• Teil: year:2016
• Teil: number:6
• Teil: elocationid:447
• Teil: pages:1-17
• Teil: extent:17
• Fussnoten: Gesehen am 10.09.2021
• Titel Quelle: Enthalten in: Remote sensing
• Ort Quelle: Basel : MDPI, 2009
• Jahr Quelle: 2016
• Band/Heft Quelle: 8(2016), 6, Artikel-ID 447, Seite 1-17
• ISSN Quelle: 2072-4292
• DOI: doi:10.3390/rs8060447
• Datenträger: Online-Ressource
• Sprache: eng
• Sach-SW: building roof
• Sach-SW: feature parameters
• Sach-SW: linear polygonal feature
• Sach-SW: point cloud registration
• K10plus-PPN: 1769997946

Abstract

This paper presents a new approach for the registration of airborne LiDAR point clouds by finding and matching corresponding linear plane features. Linear plane features are a type of common feature in an urban area and are convenient for obtaining feature parameters from point clouds. Using such linear feature parameters, the 3D rigid body coordination transformation model is adopted to register the point clouds from different trajectories. The approach is composed of three steps. In the first step, an OpenStreetMap-aided method is applied to select simply-structured roof pairs as the corresponding roof facets for the registration. In the second step, the normal vectors of the selected roof facets are calculated and input into an over-determined observation system to estimate the registration parameters. In the third step, the registration is be carried out by using these parameters. A case dataset with a two trajectory point cloud was selected to verify the proposed method. To evaluate the accuracy of the point cloud after registration, 40 checkpoints were manually selected; the results of the evaluation show that the general accuracy is 0.96 m, which is approximately 1.6 times the point cloud resolution. Furthermore, two overlap zones were selected to measure the surface-difference between the two trajectories. According to the analysis results, the average surface-distance is approximately 0.045-0.129 m.