测绘学报 ›› 2026, Vol. 55 ›› Issue (10): 1867-1877.doi: 10.11947/j.AGCS.2026.20260267

• 地图学与地理信息 • 上一篇    下一篇

基于倾斜摄影模型的低空三维航图混合分辨率体素化方法及精度约束

王继周(), 赵占杰()   

  1. 中国测绘科学研究院,北京 100036
  • 收稿日期:2026-07-10 修回日期:2026-10-14 出版日期:2026-11-16 发布日期:2026-11-16
  • 通讯作者: 赵占杰 E-mail:wangjz@casm.ac.cn;zhaozhanjie@casm.ac.cn
  • 作者简介:王继周(1976—),男,博士,研究员,研究方向为多维时空GIS、智能地图制图等。E-mail:wangjz@casm.ac.cn
  • 基金资助:
    自然资源部科技创新发展项目(102121261360000000019)

A hybrid-resolution voxelization method for oblique photogrammetric models with accuracy constraints for low-altitude 3D aeronautical charts

Jizhou Wang(), Zhanjie Zhao()   

  1. Chinese Academy of Survering and Mapping, Beijing 100036, China
  • Received:2026-07-10 Revised:2026-10-14 Online:2026-11-16 Published:2026-11-16
  • Contact: Zhanjie Zhao E-mail:wangjz@casm.ac.cn;zhaozhanjie@casm.ac.cn
  • About author:Wang Jizhou (1976—), male, PhD, researcher, majors in multi-dimensional spatio-temporal GIS, intelligent map generalization. E-mail: wangjz@casm.ac.cn
  • Supported by:
    Ministry of Natural Resources Science and Technology Innovation Development Program(102121261360000000019)

摘要:

与传统需求不同,低空经济对三维航图提出了高精度、可量测、轻量化的要求。倾斜摄影模型是构建低空三维航图最经济的数据源,但其面片表达难以直接满足低空应用需求,而体素化过程又缺乏误差传播与精度约束的理论支撑。本文的核心贡献在于将体素化重建的优化目标从“视觉保真驱动”提升为“低空安全驱动”,建立了面向低空安全的精度可控体素化理论框架,将细长结构保真、速度-制动距离等低空特殊需求纳入统一的精度约束体系。首先,建立摄影测量误差、网格插值误差与体素离散误差的可加性模型,推导出给定精度要求下体素尺度上限的闭式解;其次,构建以数据量最小为目标、以精度上限和尺度梯度为约束的变分优化模型,采用交替方向乘子法与八叉树结构进行求解;然后,提出高程加权自适应阈值方法用于剔除噪声几何,并引入锚点约束的拉普拉斯平滑以保证量测保真性;最后,选择4个典型低空场景,以12 000个随机点进行验证。结果表明,精度比均匀体素化提升了40%,碰撞检测效率比QEM简化提升近7倍,压缩率达到93%~96%,障碍物保留率为99.5%。理论分析与试验验证表明,本文方法可为构建精度可控、计算高效的体素化低空三维航图提供理论依据与实践支撑。

关键词: 低空经济, 低空三维航图, 倾斜摄影模型, 混合分辨率, 体素地图

Abstract:

Unlike traditional applications, the low-altitude economy demands 3D aeronautical charts that are highly accurate, directly measurable, and lightweight. While oblique photogrammetric models provide the most cost-effective data source for constructing such charts, their mesh-based representation falls short of operational requirements, and existing voxelization approaches lack a theoretically sound framework for error propagation and accuracy control—specifically, they fail to address the fundamental question of how voxel size should vary spatially to satisfy prescribed accuracy constraints. The core contribution of this paper is to advance voxelization from a “fidelity-driven” paradigm to a “low-altitude safety-driven” one, establishing an accuracy-controllable theoretical framework that, for the first time, incorporates low-altitude-specific requirements—including the preservation of slender structures and speed-dependent braking distance constraints—into a unified accuracy constraint system. Methodologically, we develop an additive error model that integrates photogrammetric errors, mesh interpolation errors, and voxel discretization errors, from which a closed-form solution for the upper bound of voxel size under specified accuracy requirements is derived. A variational optimization model is then formulated to minimize data volume subject to accuracy upper bounds and scale gradient constraints, solved via the alternating direction method of multipliers (ADMM) with an octree structure. Additionally, an elevation-weighted adaptive thresholding strategy is proposed to eliminate spurious geometry, while anchor-constrained Laplacian smoothing is introduced to preserve measurement fidelity. The proposed method is validated on four representative low-altitude scenarios using 12 000 randomly sampled checkpoints. Experimental results demonstrate a 40% improvement in accuracy over uniform voxelization, a nearly sevenfold increase in collision detection efficiency compared to QEM-based simplification, a compression ratio of 93%~96%, and an obstacle preservation rate of 99.5%. Both theoretical analysis and experimental evidence confirm that the proposed method offers a solid theoretical foundation and practical guidance for constructing accuracy-controllable and computationally efficient voxel-based low-altitude three-dimensional aeronautical charts.

Key words: low-altitude economy, low-altitude 3D aeronautical chart, oblique photogrammetric model, hybrid resolution, voxel map

中图分类号: