基于CVT空间变换的大场景多尺度隐式三维重建

doi:10.11947/j.AGCS.2025.20240280

摘要/Abstract

摘要：

近年来，以体渲染为基础的三维隐式表达方法在几何重建和辐射重建中取得了较好的效果，这类方法可以在重建几何结构的同时渲染接近实拍的图像。然而室外场景重建中，拍摄图像数据在空间中的不均匀分布现象，导致其重建的模型效果不佳。为解决上述问题，本文提出利用图像采样点表示图像的空间分布，通过采样点三角剖分和质心Voronoi划分迭代策略，将不均匀分布的采样点以迭代的方式变换为均匀分布，实现隐式表达分配的均匀化，从而确保隐式重建模型的质量。本文将这种对采样点进行空间变换的思想进一步嵌入不同范式的隐式表达框架中，实现室外场景的多尺度隐式重建。在室外真实场景的隐式三维重建试验结果表明，本文提出的基于质心Voronoi划分变换可以保证建筑物隐式重建精度，同时有效提升现有隐式表达多尺度的重建结果。

关键词: 摄影测量, 三维重建, 神经隐式表面, 质心Voronoi划分, 新视角合成, 体渲染

Abstract:

In recent years, volumetric rendering-based 3D implicit representation methods have achieved notable success in geometry and radiance reconstruction. These methods can simultaneously reconstruct geometric structures and render photorealistic images. However, in outdoor scene reconstruction, the uneven spatial distribution of captured images often leads to suboptimal reconstruction results. To address this issue, this paper proposes a method that represents the spatial distribution of images using sampling points. Through Delaunay triangulation and a centroidal Voronoi tessellation strategy, unevenly distributed sampling points are iteratively transformed into a uniform distribution, thereby achieving uniform allocation of the implicit representation and ensuring the quality of the implicit reconstruction. Furthermore, this concept of spatial transformation of sampling points is embedded into implicit representation frameworks of different paradigms, enabling multi-scale implicit reconstruction of outdoor scenes. Experimental results on implicit 3D reconstruction of real-world outdoor scenes demonstrate that the proposed centroidal Voronoi tessellation transformation ensures the reconstruction accuracy of building structures and significantly improves the multi-scale reconstruction results of existing implicit representation methods.

Key words: photogrammetry, 3D reconstruction, neural implicit surface, centered Voronoi tessellation, novel view synthesis, volume rendering

中图分类号:

P228

卢一鹏, 李雨昊, 王海平, 刘缘, 董震, 杨必胜. 基于CVT空间变换的大场景多尺度隐式三维重建[J]. 测绘学报, 2025, 54(4): 702-713.

Yipeng LU, Yuhao LI, Haiping WANG, Yuan LIU, Zhen DONG, Bisheng YANG. CVT space warping based multi-scale neural implicit surface reconstruction for outdoor scenes[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(4): 702-713.

图/表 12

图1

图2

图3

图4

图5

图6

表1

图7

图8

图9

图10

图11

参考文献 31

[1]	杨必胜, 陈驰, 董震. 面向智能化测绘的城市地物三维提取[J]. 测绘学报, 2022, 51(7): 1476-1484. DOI:. doi: 10.11947/j.AGCS.2022.20220183
	YANG Bisheng, CHEN Chi, DONG Zhen. 3D geospatial information extraction of urban objects for smart surveying and mapping[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(7): 1476-1484. DOI:. doi: 10.11947/j.AGCS.2022.20220183
[2]	朱庆. 三维GIS及其在智慧城市中的应用[J]. 地球信息科学学报, 2014, 16(2): 151-157.
	ZHU Qing. Full three-dimensional GIS and its key roles in smart city[J]. Journal of Geo-information Science, 2014, 16(2): 151-157.
[3]	张春森, 张会, 郭丙轩, 等. 城市场景结构感知的网格模型简化算法[J]. 测绘学报, 2020, 49(3): 334-342. DOI:. doi: 10.11947/j.AGCS.2020.20190068
	ZHANG Chunsen, ZHANG Hui, GUO Bingxuan, et al. Structure-aware simplified algorithm of mesh model for urban scene[J]. Acta Geodaetica et Cartographica Sinica, 2020, 49(3): 334-342. DOI:. doi: 10.11947/j.AGCS.2020.20190068
[4]	ZHANG Xinchang, LI Shaoying, ZHOU Qiming, et al. Logical and innovative construction of digital twin city[J]. Journal of Geodesy and Geoinformation Science, 2021, 4(4): 113-120.
[5]	CHEN Jun, LI Zhilin, LI Songnian, et al. From digitalized to intelligentized surveying and mapping: fundamental issues and research agenda[J]. Journal of Geodesy and Geoinformation Science, 2022, 5(2): 148-160.
[6]	马威, 涂强, 潘建平, 等. 桥梁实景三维高斯辐射场建模[J]. 测绘学报, 2024, 53(9): 1694-1705. DOI:. doi: 10.11947/j.AGCS.2024.20240071
	MA Wei, TU Qiang, PAN Jianping, et al. 3D Gaussian radiation field modeling for real-scene bridges[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(9): 1694-1705. DOI:. doi: 10.11947/j.AGCS.2024.20240071
[7]	杨必胜, 梁福逊, 黄荣刚. 三维激光扫描点云数据处理研究进展、挑战与趋势[J]. 测绘学报, 2017, 46(10): 1509-1516. DOI:. doi: 10.11947/j.AGCS.2017.20170351
	YANG Bisheng, LIANG Fuxun, HUANG Ronggang. Progress, challenges and perspectives of 3D LiDAR point cloud processing[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10): 1509-1516. DOI:. doi: 10.11947/j.AGCS.2017.20170351
[8]	杨必胜, 董震. 点云智能研究进展与趋势[J]. 测绘学报, 2019, 48(12): 1575-1585. DOI:. doi: 10.11947/j.AGCS.2019.20190465
	YANG Bisheng, DONG Zhen. Progress and perspective of point cloud intelligence[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(12): 1575-1585. DOI:. doi: 10.11947/j.AGCS.2019.20190465
[9]	赵煦, 周克勤, 闫利, 等. 基于激光点云的大型文物景观三维重建方法[J]. 武汉大学学报(信息科学版), 2008, 33(7): 684-687.
	ZHAO Xu, ZHOU Keqin, YAN Li, et al. 3D reconstruction method for large scale relic landscape from laser point cloud[J]. Geomatics and Information Science of Wuhan University, 2008, 33(7): 684-687.
[10]	左志强, 黄先锋, 杨冲, 等. 基于倾斜影像的城市场景隐式曲面重建[J]. 测绘通报, 2017(12): 21-28.
	ZUO Zhiqiang, HUANG Xianfeng, YANG Chong, et al. Implicit surface reconstruction of urban scenes based on oblique images[J]. Bulletin of Surveying and Mapping, 2017(12): 21-28.
[11]	张春森, 张卫龙, 郭丙轩, 等. 倾斜影像的三维纹理快速重建[J]. 测绘学报, 2015, 44(7): 782-790. DOI:. doi: 10.11947/j.AGCS.2015.20140341
	ZHANG Chunsen, ZHANG Weilong, GUO Bingxuan, et al. Rapidly 3D texture reconstruction based on oblique photography[J]. Acta Geodaetica et Cartographica Sinica, 2015, 44(7): 782-790. DOI:. doi: 10.11947/j.AGCS.2015.20140341
[12]	李明, 张卫龙, 范丁元. 城市三维重建中的自动纹理优化方法[J]. 测绘学报, 2017, 46(3): 338-345. DOI:. doi: 10.11947/j.AGCS.2017.20160467
	LI Ming, ZHANG Weilong, FAN Dingyuan. Automatic texture optimization for 3D urban reconstruction[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(3): 338-345. DOI:. doi: 10.11947/j.AGCS.2017.20160467
[13]	单杰, 李志鑫, 张文元. 大规模三维城市建模进展[J]. 测绘学报, 2019, 48(12): 1523-1541. DOI:. doi: 10.11947/j.AGCS.2019.20190471
	SHAN Jie, LI Zhixin, ZHANG Wenyuan. Recent progress in large-scale 3D city modeling[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(12): 1523-1541. DOI:. doi: 10.11947/j.AGCS.2019.20190471
[14]	KAZHDAN M, BOLITHO M, HOPPE H. Poisson surface reconstruction[C]//Proceedings of the 4th Eurographics Symposium on Geometry Processing. Goslar: Eurographics Association, 2006: 61-70.
[15]	朱艳, 颜青松, 曲英杰, 等. 影像一致性优化的视图选取策略[J]. 测绘学报, 2020, 49(11): 1463-1472. DOI:. doi: 10.11947/j.AGCS.2020.20190499
	ZHU Yan, YAN Qingsong, QU Yingjie, et al. View selection strategy for photo-consistency refinement[J]. Acta Geodaetica et Cartographica Sinica, 2020, 49(11): 1463-1472. DOI:. doi: 10.11947/j.AGCS.2020.20190499
[16]	MILDENHALL B, SRINIVASAN P P, TANCIK M, et al. NeRF: representing scenes as neural radiance fields for view synthesis[C]//Proceedings of 2020 Computer Vision. Cham: Springer, 2020: 405-421.
[17]	MÜLLER T, EVANS A, SCHIED C, et al. Instant neural graphics primitives with a multiresolution hash encoding[J]. ACM Transactions on Graphics, 2022, 41(4): 1-15.
[18]	WANG Peng, LIU Lingjie, LIU Yuan, et al. NeuS: learning neural implicit surfaces by volume rendering for multi-view reconstruction[C]//Proceedings of the 35th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2021: 27171-27183.
[19]	YARIV L, GU Jiatao, KASTEN Y, et al. Volume rendering of neural implicit surfaces[C]//Proceedings of the 35th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2021: 4805-4815.
[20]	周鑫, 王洋, 孙显, 等. 面向卫星遥感图像场景重建的神经辐射场方法综述[J]. 电子与信息学报, 2024, 46(5): 1582-1590.
	ZHOU Xin, WANG Yang, SUN Xian, et al. A review of neural radiance field approaches for scene reconstruction of satellite remote sensing imagery[J]. Journal of Electronics & Information Technology, 2024, 46(5): 1582-1590.
[21]	LORENSEN W E, CLINE H E. Marching cubes: a high resolution 3D surface construction algorithm[C]//Proceedings of the 14th Annual Conference on Computer Graphics and Interactive Techniques. New York: ACM Press, 1987: 163-169.
[22]	WANG Jiepeng, WANG Peng, LONG Xiaoxiao, et al. NeuRIS: neural reconstruction of indoor scenes using normal priors[C]//Proceedings of 2022 Computer Vision. Cham: Springer, 2022: 139-155.
[23]	GUO Haoyu, PENG Sida, LIN Haotong, et al. Neural 3D scene reconstruction with the Manhattan-world assumption[C]//Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 5501-5510
[24]	SUN Jiaming, CHEN Xi, WANG Qianqian, et al. Neural 3D reconstruction in the wild[C]//Proceedings of 2022 Special Interest Group on Computer Graphics and Interactive Techniques Conference. Vancouver: ACM Press, 2022: 1-9.
[25]	WU Tong, WANG Jiaqi, PAN Xingang, et al. Voxurf: voxel-based efficient and accurate neural surface reconstruction[EB/OL]. [2023-10-11]. https://arxiv.org/pdf/2208.12697.2023.
[26]	WANG Yiming, HAN Qin, HABERMANN M, et al. NeuS2: fast learning of neural implicit surfaces for multi-view reconstruction[C]//Proceedings of 2023 IEEE/CVF International Conference on Computer Vision. Paris: IEEE, 2023: 3272-3283.
[27]	LI Zhaoshuo, MÜLLER T, EVANS A, et al. Neuralangelo: high-fidelity neural surface reconstruction[C]//Proceedings of 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver: IEEE, 2023: 8456-8465.
[28]	ELSNER T, CZECH V, BERGER J, et al. Adaptive voronoiNeRFs[EB/OL]. [2023-06-11]. https://arxiv.org/abs/2303.16001v2.
[29]	XIAO Wenhui, CRUZ R, AHMEDT-ARISTIZABAL D, et al. Nerf director: revisiting view selection in neural volume rendering[C]//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2024: 20742-20751.
[30]	DU Qiang, FABER V, GUNZBURGER M. Centroidal Voronoi tessellations: applications and algorithms[J]. SIAM Review, 1999, 41(4): 637-676.
[31]	SCHÖNBERGER J L, FRAHM J M. Structure-from-motion revisited[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 4104-4113.

场景	类别	NeuS	NeuS Warp	Angelo	Angelo Warp
星湖楼	建筑物	0.001 10	0.001 10	0.019 79	0.016 13
星湖楼	小尺度物体	－	0.075 51	0.358 41	0.102 04
雷军楼	建筑物	0.017 71	0.017 93	0.017 56	0.017 49
雷军楼	小尺度物体	0.011 03	0.008 05	0.020 49	0.012 21
图书馆	建筑物	0.017 12	0.017 11	0.017 91	0.017 93
图书馆	小尺度物体	0.012 04	0.011 48	0.011 58	0.010 95
物理学院	建筑物	0.018 31	0.018 73	0.017 01	0.017 01
物理学院	小尺度物体	0.007 37	0.004 97	0.009 67	0.005 30