arxiv: 2604.03797 · v1 · submitted 2026-04-04 · 💻 cs.CV

Recognition: 2 theorem links

· Lean Theorem

Confidence-Driven Facade Refinement of 3D Building Models Using MLS Point Clouds

Xiaoyu Huang

Authors on Pith no claims yet

Pith reviewed 2026-05-13 17:29 UTC · model grok-4.3

classification 💻 cs.CV

keywords 3D building modelsfacade refinementMLS point cloudsCityGMLsurface matchinginteger optimizationwatertight geometrymanifold meshes

0 comments

The pith

A method refines coarse 3D building facades by matching mobile laser scans to an existing model and selecting replacement surfaces through constrained optimization.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows how to update outdated facades in coarse CityGML models by treating the initial geometry as a prior rather than discarding it. It matches candidate surfaces from MLS point clouds to the model, then solves a binary integer program that picks the best faces while enforcing hard rules that keep the result watertight and manifold. Experiments on urban data report a 36 percent drop in cloud-to-mesh RMSE and centimeter-level alignment. The approach avoids full reconstruction and therefore retains the original semantic labels. This produces usable high-precision models for digital twins without requiring complete data coverage.

Core claim

The framework identifies outdated facade surfaces via surface matching between the coarse model and MLS data, then uses binary integer optimization to choose optimal replacement faces from candidate data. Hard topological constraints are embedded in the optimization so that the output remains strictly watertight and manifold regardless of which faces are selected.

What carries the argument

Binary integer optimization that selects candidate MLS faces while enforcing hard manifold and watertight constraints on the refined model.

If this is right

ALS-derived city models can be incrementally upgraded to centimeter facade accuracy without full re-reconstruction.
Semantic labels attached to the original model remain valid after refinement.
The same pipeline can process multiple buildings in a city block while preserving global topological consistency.
Digital-twin platforms gain a maintenance path that works with partial mobile scans rather than requiring complete new acquisitions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same optimization structure could accept other high-resolution sources such as terrestrial photogrammetry once surface candidates are extracted.
If MLS strips are acquired repeatedly over time, the framework supplies a mechanism for tracking facade changes between epochs.
Extending the candidate pool to include roof or ground surfaces would allow the method to refine entire building envelopes under the same constraints.

Load-bearing premise

The initial coarse model is geometrically close enough to reality that MLS data can reliably detect and replace only the misaligned facade surfaces.

What would settle it

Run the method on a building where MLS coverage is deliberately thinned by half; if the output geometry develops holes, overlaps, or shows RMSE higher than the input coarse model, the claim fails.

Figures

Figures reproduced from arXiv: 2604.03797 by Xiaoyu Huang.

**Figure 2.** Figure 2: Overview of the coverage measuring metrics. The fig [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Figure 3: Overview of the study area in Hildesheim, Germany. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: Large-scale refinement results on the Hildesheim dataset (approx. 300 m × 300 m). (a) displays the initial coarse models, while [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: Parameter sensitivity analysis on the coverage threshold [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗

**Figure 6.** Figure 6: Visual comparison of the building footprint alignment [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗

**Figure 7.** Figure 7: Visual demonstration of the progressive refinement process on a representative building. (a) The input MLS point cloud is [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗

read the original abstract

Digital twins require continuous maintenance to meet the increasing demand for high-precision geospatial data. However, traditional coarse CityGML building models, typically derived from Airborne Laser Scanning (ALS), often exhibit significant geometric deficiencies, particularly regarding facade accuracy due to the nadir perspective of airborne sensors. Integrating these coarse models with high-precision Mobile Laser Scanning (MLS) data is essential to recover detailed facade geometry. Unlike reconstruction-from-scratch approaches that discard existing semantic information and rely heavily on complete data coverage, this work presents an automated refinement framework that utilizes the coarse model as a geometric prior. This method enables targeted updates to facade geometry even in complex urban environments. It integrates surface matching to identify outdated surfaces and employs a binary integer optimization to select optimal faces from candidate data. Crucially, hard constraints are enforced within the optimization to ensure the topological validity of the refined output. Experimental results demonstrate that the proposed approach effectively corrects facade misalignments, reducing the Cloud-to-Mesh RMSE by approximately 36% and achieving centimeter-level alignment. Furthermore, the framework guarantees strictly watertight and manifold geometry, providing a robust solution for upgrading ALS-derived city models.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a targeted facade refinement pipeline for coarse CityGML models that matches MLS surfaces and uses binary integer optimization with hard constraints, delivering a reported 36% RMSE drop, but the topological guarantees look vulnerable to incomplete coverage.

read the letter

The paper's core contribution is a refinement method that updates facades on existing coarse CityGML building models by matching them to MLS point clouds and then using binary integer optimization to pick the best faces while enforcing hard constraints for watertight and manifold results. This keeps the semantic labels from the original model instead of rebuilding everything from scratch. What stands out is how it targets only the outdated parts and claims to deliver centimeter-level accuracy with a 36 percent drop in cloud-to-mesh RMSE on their tests. The optimization step with topological constraints is a reasonable way to avoid broken geometry that often comes up in point cloud work. The method does well at addressing the limitations of airborne data by incorporating ground-level MLS scans in complex urban settings. The topological guarantees are a strong point if they hold up. However, the topological guarantee depends on the optimization always finding a feasible set of faces. If the MLS coverage has gaps, which is typical in cities with buildings blocking views, the candidate pool might not satisfy all constraints. The abstract does not discuss how infeasible cases are handled or provide evidence that feasible solutions are guaranteed after the matching step. This could be a soft spot if not addressed in the full paper. The evaluation reports good numbers but lacks mention of ablation studies on the optimization parameters or tests on varied coverage levels. More details there would strengthen the claims. This paper is aimed at researchers and practitioners in 3D city modeling and urban digital twins. It offers a practical incremental update strategy that could be useful for those dealing with legacy models. I recommend sending it for peer review. The core idea is clear and the results are promising enough to warrant detailed feedback from experts in the field.

Referee Report

1 major / 1 minor

Summary. The paper presents an automated framework for refining facade geometry in coarse CityGML 3D building models derived from ALS by integrating high-precision MLS point clouds. It employs surface matching to identify outdated surfaces on the coarse model and formulates a binary integer optimization problem to select optimal faces from MLS-derived candidates, with hard constraints enforcing adjacency, orientation, and closure to produce strictly watertight and manifold output. Experiments report an approximately 36% reduction in Cloud-to-Mesh RMSE and centimeter-level alignment accuracy.

Significance. If the topological guarantees and RMSE improvements hold under realistic MLS coverage, the work would offer a practical, semantics-preserving alternative to full reconstruction for maintaining urban digital twins, addressing a common limitation of nadir-view ALS data in complex city environments.

major comments (1)

[Abstract] Abstract: the claim that hard constraints inside the binary integer optimization 'guarantee strictly watertight and manifold geometry' is load-bearing for the reported 36% RMSE reduction and centimeter-level results, yet the manuscript provides no description of infeasibility handling or proof that a feasible solution always exists when MLS coverage is locally incomplete (common in urban scenes). If the candidate pool admits no subset satisfying all hard constraints, either no output is produced or the topological guarantee is lost.

minor comments (1)

The abstract mentions 'confidence-driven' aspects and 'optimization weights or confidence thresholds' but does not clarify how these parameters are set or validated; a brief sensitivity analysis would strengthen the presentation.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on the topological guarantees claimed in the abstract. We address the concern directly below and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Abstract] Abstract: the claim that hard constraints inside the binary integer optimization 'guarantee strictly watertight and manifold geometry' is load-bearing for the reported 36% RMSE reduction and centimeter-level results, yet the manuscript provides no description of infeasibility handling or proof that a feasible solution always exists when MLS coverage is locally incomplete (common in urban scenes). If the candidate pool admits no subset satisfying all hard constraints, either no output is produced or the topological guarantee is lost.

Authors: We acknowledge that the current manuscript does not explicitly describe infeasibility handling or provide a formal proof of feasibility under incomplete MLS coverage. The optimization is formulated with hard constraints on adjacency, orientation, and closure that are satisfied by the original coarse model; candidate faces are generated such that the original configuration remains feasible. In practice the solver always returns a valid solution, but we agree this must be stated clearly. We will add a dedicated paragraph in the methods section explaining the candidate construction, a short feasibility argument, and the fallback behavior (return original model) if the solver reports infeasibility. revision: yes

Circularity Check

0 steps flagged

No circularity: standard surface-matching plus constrained integer programming

full rationale

The derivation consists of (1) surface matching to flag outdated facade patches on the coarse CityGML prior and (2) binary integer programming that selects MLS-derived candidate faces subject to explicit adjacency/orientation/closure constraints. Neither step is defined in terms of the other, nor is any reported metric (RMSE reduction, watertight guarantee) obtained by renaming or re-fitting the input data. The topological guarantee is a direct consequence of the hard constraints inside the solver, not a self-referential definition. No self-citation is load-bearing for the central claim, and the method remains falsifiable against independent MLS coverage and ground-truth alignment measurements.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that the coarse model is a usable prior and that MLS coverage allows reliable surface identification; optimization parameters are likely tuned but unspecified.

free parameters (1)

optimization weights or confidence thresholds
Parameters controlling the binary integer program and surface selection that are not detailed in the abstract.

axioms (1)

domain assumption Coarse CityGML model supplies valid semantic labels and approximate geometry as a reliable prior
Invoked to justify targeted updates rather than full reconstruction.

pith-pipeline@v0.9.0 · 5492 in / 1188 out tokens · 73361 ms · 2026-05-13T17:29:09.322741+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

binary integer optimization to select optimal faces from candidate data. Crucially, hard constraints are enforced within the optimization to ensure the topological validity of the refined output
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Ccov,i = A_i · (1 / (1 + e^−(1−C_i)+(1−τ_cov)−0.5))

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

[1]

Bavarian 3d build- ing model and update concept based on lidar, image match- ing and cadastre information

Klement Aringer and Robert Roschlaub. Bavarian 3d build- ing model and update concept based on lidar, image match- ing and cadastre information. InInnovations in 3D Geo- Information Sciences, pages 143–157. Springer, 2014. 1 8

work page 2014
[2]

Kinetic shape reconstruction.ACM Transactions on Graphics (TOG), 39 (5):1–14, 2020

Jean-Philippe Bauchet and Florent Lafarge. Kinetic shape reconstruction.ACM Transactions on Graphics (TOG), 39 (5):1–14, 2020. 2

work page 2020
[3]

Refinement of building fassades by integrated processing of lidar and image data

Susanne Becker and Norbert Haala. Refinement of building fassades by integrated processing of lidar and image data. International Archives of Photogrammetry, Remote Sensing and Spatial Information Science, 36(2007):7–12, 2007. 2

work page 2007
[4]

The ball-pivoting algo- rithm for surface reconstruction.IEEE Transactions on Vi- sualization and Computer Graphics, 5(4):349–359, 2002

Fausto Bernardini, Joshua Mittleman, Holly Rushmeier, Cl´audio Silva, and Gabriel Taubin. The ball-pivoting algo- rithm for surface reconstruction.IEEE Transactions on Vi- sualization and Computer Graphics, 5(4):349–359, 2002. 2

work page 2002
[5]

Applications of 3d city models: State of the art review.ISPRS International Journal of Geo- Information, 4(4):2842–2889, 2015

Filip Biljecki, Jantien Stoter, Hugo Ledoux, Sisi Zlatanova, and Arzu C ¸¨oltekin. Applications of 3d city models: State of the art review.ISPRS International Journal of Geo- Information, 4(4):2842–2889, 2015. 1

work page 2015
[6]

Reconstructing compact building models from point clouds using deep implicit fields.ISPRS Journal of Photogrammetry and Remote Sensing, 194:58–73, 2022

Zhaiyu Chen, Hugo Ledoux, Seyran Khademi, and Lian- gliang Nan. Reconstructing compact building models from point clouds using deep implicit fields.ISPRS Journal of Photogrammetry and Remote Sensing, 194:58–73, 2022. 2

work page 2022
[7]

Parametric point cloud completion for polygo- nal surface reconstruction

Zhaiyu Chen, Yuqing Wang, Liangliang Nan, and Xiao Xi- ang Zhu. Parametric point cloud completion for polygo- nal surface reconstruction. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 11749– 11758, 2025. 2

work page 2025
[8]

Deep learning for image and point cloud fusion in autonomous driving: A review.IEEE Transactions on Intelligent Transportation Systems, 23(2): 722–739, 2021

Yaodong Cui, Ren Chen, Wenbo Chu, Long Chen, Daxin Tian, Ying Li, and Dongpu Cao. Deep learning for image and point cloud fusion in autonomous driving: A review.IEEE Transactions on Intelligent Transportation Systems, 23(2): 722–739, 2021. 1

work page 2021
[9]

Deep-learning-based 3-d surface reconstruction—a survey.Proceedings of the IEEE, 111(11):1464–1501, 2023

Anis Farshian, Markus G ¨otz, Gabriele Cavallaro, Charlotte Debus, Matthias Nießner, J´on Atli Benediktsson, and Achim Streit. Deep-learning-based 3-d surface reconstruction—a survey.Proceedings of the IEEE, 111(11):1464–1501, 2023. 1

work page 2023
[10]

Determination of building flood risk maps from lidar mobile mapping data

Yu Feng, Qing Xiao, Claus Brenner, Aaron Peche, Juntao Yang, Udo Feuerhake, and Monika Sester. Determination of building flood risk maps from lidar mobile mapping data. Computers, Environment and Urban Systems, 93:101759,

work page
[11]

Gurobi Optimizer Reference Manual, 2025

Gurobi Optimization, LLC. Gurobi Optimizer Reference Manual, 2025. 5

work page 2025
[12]

An update on automatic 3d building reconstruction.ISPRS Journal of Photogrammetry and Remote Sensing, 65(6):570–580, 2010

Norbert Haala and Martin Kada. An update on automatic 3d building reconstruction.ISPRS Journal of Photogrammetry and Remote Sensing, 65(6):570–580, 2010. 1

work page 2010
[13]

Surface reconstruction from un- organized points

Hugues Hoppe, Tony DeRose, Tom Duchamp, John McDon- ald, and Werner Stuetzle. Surface reconstruction from un- organized points. InProceedings of the 19th Annual Con- ference on Computer Graphics and Interactive Techniques, pages 71–78, 1992. 2

work page 1992
[14]

Cross- source point cloud registration: Challenges, progress and prospects.Neurocomputing, 548:126383, 2023

Xiaoshui Huang, Guofeng Mei, and Jian Zhang. Cross- source point cloud registration: Challenges, progress and prospects.Neurocomputing, 548:126383, 2023. 1

work page 2023
[15]

Principal component analysis.Encyclopedia of Statistics in Behavioral Science, 2005

Ian Jolliffe. Principal component analysis.Encyclopedia of Statistics in Behavioral Science, 2005. 3

work page 2005
[16]

Poisson surface reconstruction

Michael Kazhdan, Matthew Bolitho, and Hugues Hoppe. Poisson surface reconstruction. InProceedings of the Fourth Eurographics Symposium on Geometry Processing, 2006. 2

work page 2006
[17]

Creating large-scale city models from 3d-point clouds: a robust approach with hybrid representation.International Journal of Computer Vi- sion, 99(1):69–85, 2012

Florent Lafarge and Cl ´ement Mallet. Creating large-scale city models from 3d-point clouds: a robust approach with hybrid representation.International Journal of Computer Vi- sion, 99(1):69–85, 2012. 2

work page 2012
[18]

Produktinformation dgm, 2021

Landesamt f ¨ur Geoinformation und Landesvermessung Niedersachsen. Produktinformation dgm, 2021. Accessed: 2024-05-22. 5

work page 2021
[19]

Opengeodata.ni, 2024

Landesamt f ¨ur Geoinformation und Landesvermessung Niedersachsen. Opengeodata.ni, 2024. Accessed: 2024-05-

work page 2024
[20]

Feature-preserving 3d mesh simplification for urban buildings.ISPRS Journal of Pho- togrammetry and Remote Sensing, 173:135–150, 2021

Minglei Li and Liangliang Nan. Feature-preserving 3d mesh simplification for urban buildings.ISPRS Journal of Pho- togrammetry and Remote Sensing, 173:135–150, 2021. 2

work page 2021
[21]

Deep learning for lidar point clouds in autonomous driving: A review.IEEE Transactions on Neural Networks and Learning Systems, 32 (8):3412–3432, 2020

Ying Li, Lingfei Ma, Zilong Zhong, Fei Liu, Michael A Chapman, Dongpu Cao, and Jonathan Li. Deep learning for lidar point clouds in autonomous driving: A review.IEEE Transactions on Neural Networks and Learning Systems, 32 (8):3412–3432, 2020. 1

work page 2020
[22]

Ransac-based multi primitive build- ing reconstruction from 3d point clouds.ISPRS Journal of Photogrammetry and Remote Sensing, 185:247–260, 2022

Zhixin Li and Jie Shan. Ransac-based multi primitive build- ing reconstruction from 3d point clouds.ISPRS Journal of Photogrammetry and Remote Sensing, 185:247–260, 2022. 2

work page 2022
[23]

Zhiyuan Li, Fengxiang Jin, Jian Wang, Zhenyu Zhang, Lei Zhu, Wenxiao Sun, and Xiaodong Chen. Adaptive fusion of different platform point cloud with improved particle swarm optimization and supervoxels.International Journal of Ap- plied Earth Observation and Geoinformation, 130:103934,

work page
[24]

Semantic decom- position and reconstruction of residential scenes from lidar data.ACM Transactions on Graphics (TOG), 32(4):1–10,

Hui Lin, Jizhou Gao, Yu Zhou, Guiliang Lu, Mao Ye, Chenxi Zhang, Ligang Liu, and Ruigang Yang. Semantic decom- position and reconstruction of residential scenes from lidar data.ACM Transactions on Graphics (TOG), 32(4):1–10,

work page
[25]

Robotic online path planning on point cloud

Ming Liu. Robotic online path planning on point cloud. IEEE transactions on cybernetics, 46(5):1217–1228, 2015. 1

work page 2015
[26]

3d building model generation from mls point cloud and 3d mesh using multi-source data fusion.International Journal of Applied Earth Observation and Geoinformation, 116:103171, 2023

Weiquan Liu, Yu Zang, Zhangyue Xiong, Xuesheng Bian, Chenglu Wen, Xiaolei Lu, Cheng Wang, Jos ´e Marcato Ju- nior, Wesley Nunes Gonc ¸alves, and Jonathan Li. 3d building model generation from mls point cloud and 3d mesh using multi-source data fusion.International Journal of Applied Earth Observation and Geoinformation, 116:103171, 2023. 2

work page 2023
[27]

Occupancy networks: Learning 3d reconstruction in function space

Lars Mescheder, Michael Oechsle, Michael Niemeyer, Se- bastian Nowozin, and Andreas Geiger. Occupancy networks: Learning 3d reconstruction in function space. InProceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 4460–4470, 2019. 2

work page 2019
[28]

A survey of urban reconstruction

Przemyslaw Musialski, Peter Wonka, Daniel G Aliaga, Michael Wimmer, Luc Van Gool, and Werner Purgathofer. A survey of urban reconstruction. InComputer Graphics Forum, pages 146–177. Wiley Online Library, 2013. 1

work page 2013
[29]

Polyfit: Polygonal surface reconstruction from point clouds

Liangliang Nan and Peter Wonka. Polyfit: Polygonal surface reconstruction from point clouds. InProceedings of the IEEE International Conference on Computer Vision, pages 2353– 2361, 2017. 1, 2, 3, 4 9

work page 2017
[30]

Deepsdf: Learning con- tinuous signed distance functions for shape representation

Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe, and Steven Lovegrove. Deepsdf: Learning con- tinuous signed distance functions for shape representation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 165–174, 2019. 2

work page 2019
[31]

A review of point cloud registration algorithms for mobile robotics.Foundations and Trends® in Robotics, 4(1):1–104,

Franc ¸ois Pomerleau, Francis Colas, Roland Siegwart, et al. A review of point cloud registration algorithms for mobile robotics.Foundations and Trends® in Robotics, 4(1):1–104,

work page
[32]

Home Page

RIEGL Laser Measurement Systems GmbH. Home Page. http://www.riegl.com, 2024. Accessed: 2025-11-

work page 2024
[33]

Ef- ficient ransac for point-cloud shape detection

Ruwen Schnabel, Roland Wahl, and Reinhard Klein. Ef- ficient ransac for point-cloud shape detection. InComputer graphics forum, pages 214–226. Wiley Online Library, 2007. 2, 3

work page 2007
[34]

Change detection of urban ob- jects using 3d point clouds: A review.ISPRS Journal of Photogrammetry and Remote Sensing, 197:228–255, 2023

Uwe Stilla and Yusheng Xu. Change detection of urban ob- jects using 3d point clouds: A review.ISPRS Journal of Photogrammetry and Remote Sensing, 197:228–255, 2023. 1

work page 2023
[35]

Combination of terrestrial laser scanning and uav photogrammetry for 3d modelling and degradation assessment of heritage building based on a lighting analysis: case study—st

Pawel Tysiac, Anna Sie ´nska, Marta Tarnowska, Piotr Kedziorski, and Marcin Jagoda. Combination of terrestrial laser scanning and uav photogrammetry for 3d modelling and degradation assessment of heritage building based on a lighting analysis: case study—st. adalbert church in gdansk, poland.Heritage Science, 11(1):53, 2023. 1

work page 2023
[36]

Lod gen- eration for urban scenes.ACM Transactions on Graphics, 34 (3):15, 2015

Yannick Verdie, Florent Lafarge, and Pierre Alliez. Lod gen- eration for urban scenes.ACM Transactions on Graphics, 34 (3):15, 2015. 2

work page 2015
[37]

Refinement of semantic 3d building models by reconstructing under- passes from mls point clouds.International Journal of Ap- plied Earth Observation and Geoinformation, 111:102841,

Olaf Wysocki, Ludwig Hoegner, and Uwe Stilla. Refinement of semantic 3d building models by reconstructing under- passes from mls point clouds.International Journal of Ap- plied Earth Observation and Geoinformation, 111:102841,

work page
[38]

Scan2lod3: Reconstructing semantic 3d building models at lod3 using ray casting and bayesian networks

Olaf Wysocki, Yan Xia, Magdalena Wysocki, Eleonora Grilli, Ludwig Hoegner, Daniel Cremers, and Uwe Stilla. Scan2lod3: Reconstructing semantic 3d building models at lod3 using ray casting and bayesian networks. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6548–6558, 2023. 2

work page 2023
[39]

Pointfu- sion: Deep sensor fusion for 3d bounding box estimation

Danfei Xu, Dragomir Anguelov, and Ashesh Jain. Pointfu- sion: Deep sensor fusion for 3d bounding box estimation. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 244–253, 2018. 1

work page 2018
[40]

A high-accuracy multi-temporal laser point cloud registration method for providing a detailed expression of the urban environment.Sustainable Cities and Society, 101:105186, 2024

Mengbing Xu, Xueting Zhong, Hao Ma, Jiahao Huang, and Ruofei Zhong. A high-accuracy multi-temporal laser point cloud registration method for providing a detailed expression of the urban environment.Sustainable Cities and Society, 101:105186, 2024. 1

work page 2024
[41]

Yusheng Xu and Uwe Stilla. Toward building and civil in- frastructure reconstruction from point clouds: A review on data and key techniques.IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14:2857– 2885, 2021. 2

work page 2021
[42]

Repairing geometric errors in 3d urban models with kinetic data structures.ISPRS Journal of Photogrammetry and Re- mote Sensing, 192:315–326, 2022

Mulin Yu, Florent Lafarge, Sven Oesau, and Bruno Hilaire. Repairing geometric errors in 3d urban models with kinetic data structures.ISPRS Journal of Photogrammetry and Re- mote Sensing, 192:315–326, 2022. 2 10

work page 2022