Scan2LoD3: Reconstructing semantic 3D building models at LoD3 using ray casting and Bayesian networks

Wysocki, Olaf; Xia, Yan; Magdalena, Wysocki,; Grilli, Eleonora; Hoegner, L.; Cremers, Daniel; Stilla, Uwe

doi:10.1109/cvprw59228.2023.00696

Cited by 6 publications

(2 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While numerous studies have focused on the automatic creation of 3D buildings in CityGML [23][24][25][26][27], there remains a scarcity of any standard workflow for acquiring, processing, and presenting implementations for higher levels of detail, specifically Level of Detail 3 (LOD3) [28]. Most LOD3 models are derived from simplified BIM models [29,30] or as an extension of already existing LOD2 models [31,32]. This study contributes a semi-automated yet complete workflow for disseminating point cloud information into BIM, which later translates into a 3D GIS schema, culminating in the development of a CityGML model up to LOD3 from point clouds.…”

Section: Introductionmentioning

confidence: 99%

Low-Cost Data, High-Quality Models: A Semi-Automated Approach to LOD3 Creation

Harshit,

Chaurasia,

Zlatanova

et al. 2024

IJGI

View full text Add to dashboard Cite

In the dynamic realm of digital twin modeling, where advancements are swiftly unfolding, users now possess the unprecedented ability to capture and generate geospatial data in real time. This article delves into a critical exploration of this landscape by presenting a meticulously devised workflow tailored for the creation of Level of Detail 3 (LOD3) models. Our research methodology capitalizes on the integration of Apple LiDAR technology alongside photogrammetric point clouds acquired from Unmanned Aerial Vehicles (UAVs). The proposed process unfolds with the transformation of point cloud data into Industry Foundation Classes (IFC) models, which are subsequently refined into LOD3 Geographic Information System (GIS) models leveraging the Feature Manipulation Engine (FME) workbench 2022.1.2. This orchestrated synergy among Apple LiDAR, UAV-derived photogrammetric point clouds, and the transformative capabilities of the FME culminates in the development of precise LOD3 GIS models. Our proposed workflow revolutionizes this landscape by integrating multi-source point clouds, imbuing them with accurate semantics derived from IFC models, and culminating in the creation of valid CityGML LOD3 buildings through sophisticated 3D geometric operations. The implications of this technical innovation are profound. Firstly, it elevates the capacity to produce intricate infrastructure models, unlocking new vistas for modeling digital twins. Secondly, it extends the horizons of GIS applications by seamlessly integrating enriched Building Information Modeling (BIM) components, thereby enhancing decision-making processes and facilitating more comprehensive spatial analyses.

show abstract

Section: Introductionmentioning

confidence: 99%

Low-Cost Data, High-Quality Models: A Semi-Automated Approach to LOD3 Creation

Harshit,

Chaurasia,

Zlatanova

et al. 2024

IJGI

View full text Add to dashboard Cite

show abstract

“…Since rather recently, 3D building models at LOD2 have been widely generated [20][21][22] with the use of ALS data, satellite images, and deep learning algorithms [20,23,24]. TLS and ALS data are used to develop LOD3 models of buildings that play an important role in urban space [25][26][27][28]. Novel solutions for automating the generation of LOD3 building models have been reported in the literature [29,30].…”

Section: Introductionmentioning

confidence: 99%

Modeling the Geometry of Tree Trunks Using LiDAR Data

Tarsha Kurdi,

Gharineiat,

Lewandowicz

et al. 2024

Forests

View full text Add to dashboard Cite

The effective development of digital twins of real-world objects requires sophisticated data collection techniques and algorithms for the automated modeling of individual objects. In City Information Modeling (CIM) systems, individual buildings can be modeled automatically at the second Level of Detail or LOD2. Similarly, for Tree Information Modeling (TIM) and building Forest Digital Twins (FDT), automated solutions for the 3D modeling of individual trees at different levels of detail are required. The existing algorithms support the automated modeling of trees by generating models of the canopy and the lower part of the trunk. Our argument for this work is that the structure of tree trunk and branches is as important as canopy shape. As such, the aim of the research is to develop an algorithm for automatically modeling tree trunks based on data from point clouds obtained through laser scanning. Aiming to generate 3D models of tree trunks, the suggested approach starts with extracting the trunk point cloud, which is then segmented into single stems. Subsets of point clouds, representing individual branches, are measured using Airborne Laser Scanning (ALS) and Terrestrial Laser Scanning (TLS). Trunks and branches are generated by fitting cylinders to the layered subsets of the point cloud. The individual stems are modeled by a structure of slices. The accuracy of the model is calculated by determining the fitness of cylinders to the point cloud. Despite the huge variation in trunk geometric forms, the proposed modeling approach can gain an accuracy of better than 4 cm in the constructed tree trunk models. As the developed tree models are represented in a matrix format, the solution enables automatic comparisons of tree elements over time, which is necessary for monitoring changes in forest stands. Due to the existence of large variations in tree trunk geometry, the performance of the proposed modeling approach deserves further investigation on its generality to other types of trees in multiple areas.

show abstract