広がる空間情報の整備・公開～都市モデルからUAVによる点群まで～

抄録

The techniques for map creation are evolving rapidly, driven by technological advancements and the growing need for spatial information data in various fields. Since the concept of ‘maps’ varies significantly depending on user needs and context, the maps considered in this paper focus on spatial information, including large-scale 3D models for urban planning and high accuracy detailed spatial information used at construction sites of civil engineering. Traditional principles such as triangulation remain foundational, but rapid advancements in GNSS positioning, photogrammetry, and laser scanning have significantly improved spatial and temporal resolution. As a result, outputs have progressed from analog line drawings to digital vector and raster data, orthophotos, and dense 3D point clouds, enabling end‑to‑end 3D workflows across planning, design, construction, inspection, and asset management.

With Japan facing population decline and labor shortages, digital transformation (DX) and digital‑twin technologies are becoming essential in construction and urban management. A key initiative highlighted is the Ministry of Land, Infrastructure, Transport and Tourism(MLIT)’s Project PLATEAU, launched in 2020 to build, standardize, and release 3D urban models as open data. Using aerial photography and laser scanning, local governments generate detailed 3D city models. These models follow the CityGML standard, an international XML‑based format that classifies urban objects by type and level of detail (LOD0–LOD4). Japan also developed an extension called i‑UR (i‑Urban Regeneration), adding attributes for quantitative urban analysis, temporal updates, and more granular LODs (e.g., LOD2.1–2.2, LOD3.1–3.3). As of 2025, more than 400 cities have released their 3D models through the G-Spatial Information Center. Shizuoka Prefecture has been advancing pioneering activities to maintain and publish 3D point cloud data under the concept “VIRTUAL SHIZUOKA” which leverages 3D point cloud data to solve societal issues. From 2019 to 2025, Shizuoka Prefecture performed laser scanning from airborne and mobile platforms across all its land areas, acquiring over 30TB of point cloud data. This dataset is publicly available under an open license and supports disaster prevention, infrastructure maintenance, autonomous driving, and tourism. The utility of open 3D data was demonstrated during the 2021 Atami debris‑flow disaster, where rapid comparison point cloud from multiple time periods to assess land changes in the disaster area allowed researchers to estimate sediment deposition within one day of the disaster.

To raise productivity, MLIT’s i‑Construction and the upgraded i‑Construction 2.0 integrate ICT across construction sites, promoting consistent 3D data use from survey through maintenance to reduce rework, enable mechanization, and improve efficiency. UAV‑based acquisition has become mainstream: UAV photogrammetry (SfM/MVS) generates high‑density point clouds from highly overlapped imagery but demands careful quality and data management; UAV Lidar provides faster processing and retrieves terrain under vegetation, though discrete sampling limits edge capture without re‑measurement. Recent technical innovations include handheld/backpack LidarSLAM (Lidar Simultaneous Localization And Mapping) systems for on‑site 3D acquisition. Furthermore, emerging image-based reconstruction techniques such as 3D Gaussian splatting enable realistic rendering of challenging materials like specular reflections and transparency in photogrammetry, expanding their application fields.

Finally, this paper summarizes that the maintenance and open publication of spatial information, achieved through initiatives such as the MLIT’s Project PLATEAU and Shizuoka Prefecture’s VIRTUAL SHIZUOKA, along with the utilization of UAV based point clouds at construction sites, contributes to solving societal issues.