SRTalk to our team→
Dense georeferenced urban point cloud showing buildings, streets and vegetation inside the Belgrade survey area.
3D REALITY CAPTURE · URBAN BUILDING STOCK · HORIZON EUROPE

CIRC-BOOST Belgrade — 3D Building Survey for Circular-Construction Research

Reality capture → multisensor point-cloud integration → classification → validated CityGML LoD2 building models.

CIRC-BOOST Belgrade integrates UAV LiDAR, mobile mapping, GNSS control, point-cloud classification and CityGML LoD2 modelling for urban building-stock research.

LocationPalilula municipality, Belgrade
ClientUniversity of Belgrade, Faculty of Civil Engineering
Period2023-12 — 2025-02
Project context

A dense Belgrade neighbourhood had to become a reliable 3D research foundation

Within the wider CIRC-BOOST programme, MapSoft's documented role was the specialist geospatial work package: capture complementary roof and street-level geometry, integrate independent LiDAR sources and deliver a controlled LoD2 spatial foundation for downstream building-stock research.

717 buildingsCityGML LoD2 buildings
38 missionsUAV flight missions
5 missionsMobile mapping missions
0.02 m3d positioning standard deviation
Map of the Palilula study area used to organize the CIRC-BOOST urban 3D survey.
Belgrade project area and building footprint context
MapSoft scope

Subcontracted geomatics service provider to the University of Belgrade Faculty of Civil Engineering: UAV LiDAR, mobile mapping, GNSS control, georeferencing, point-cloud processing/classification, QA/QC and CityGML LoD2 production.

01Plan UAV blocks, GCP distribution and mobile routes
02Acquire aerial and street-level LiDAR/image data
03Process trajectories and raw sensor data
04Align aerial and mobile point clouds
05Classify points and remove noise/misclassification
06Generate, inspect and correct LoD2 geometry
07Deliver classified LAZ and CityGML LoD2
Multisensor acquisition

Roofs from the air. Façades and street detail from the ground.

The acquisition strategy combined UAV LiDAR with vehicle-based mobile mapping and GNSS control. Six UAV blocks, 38 UAV missions and five mobile-mapping missions were organized to cover a dense urban area where no single viewpoint could provide the required geometry.

Mobile mapping in operation during the Belgrade survey.
MapSoft survey vehicle equipped for street-level mobile mapping in the dense urban project area.
MapSoft mobile mapping system operating in Belgrade
DJI Matrice platform prepared for urban UAV LiDAR acquisition.
UAV LiDAR system prepared for acquisition
Field GNSS measurement used for georeferencing and independent point-cloud accuracy checks.
GNSS ground-control measurement
Mission planning
Area-of-interest division into six UAV LiDAR acquisition blocks.
Six UAV LiDAR acquisition blocks
Street network and planned mobile mapping routes designed for complete coverage and overlapping passes.
Mobile routes planned for complete street coverage and overlapping passes
Point-cloud production

Independent sensor streams were processed, georeferenced and aligned into one urban 3D dataset

Aerial LiDAR was processed from raw sensor data, mobile trajectories were solved with GNSS/INS processing, and the resulting point clouds were aligned and classified before 3D building production.

Dense urban aerial point cloud in the production environment during raw LiDAR processing.
Aerial LiDAR point cloud in DJI Terra
Urban UAV point cloud after RGB colourization, preserving building and vegetation detail.
RGB-colourized UAV point cloud
Street-level façade and urban detail captured by mobile LiDAR after georeferencing.
Georeferenced mobile point cloud
Integrated 3D data

The combined cloud carries both roof geometry and street-level detail

The strongest project evidence is not a single sensor result, but the complementary geometry produced by aerial and mobile mapping together.

RGB visualization of the integrated urban point-cloud dataset.
Integrated RGB point-cloud deliverable
Urban classified cloud differentiating building, ground and other project classes.
Classified point cloud
LoD2 building modelling

From measured reality to structured CityGML geometry

Building geometry was generated from the classified point cloud, manually reviewed and topologically checked. The documented MapSoft deliverable is CityGML LoD2 — not the later LoD3 or Digital Twin work performed elsewhere in the wider programme.

Production view showing the transition from 2D building footprints to structured 3D LoD2 geometry.
From 2D footprints to LoD2 building geometry
Final structured 3D building output showing roof, wall and base geometry for the delivered LoD2 building stock.
CityGML LoD2 urban building model
LAS

Georeferenced aerial LiDAR point cloud

UAV LiDAR point cloud processed and georeferenced for the project area.

Point cloud

Georeferenced mobile LiDAR point cloud

Street-level mobile LiDAR aligned across passes and matched with the aerial cloud.

LAZ

Classified point cloud

Complete area; automatic/manual classification, noise removal, classes 1/2/6, divided into 16 delivery blocks.

16 blocks
CityGML

CityGML LoD2 building model

717 buildings with base surfaces, walls and roofs; automatic generation followed by manual geometry/topology correction.

717 buildings
Documentation

Technical and QA documentation

Technical report covering acquisition, processing, equipment, georeferencing, QA/QC and final products.

Quality assurance

Trajectory and independent height checks provide quantitative proof

QA combined trajectory monitoring, independent GCP comparisons and manual/topological checks of the final LoD2 building geometry, including centimetre-level control statistics.

Position, attitude, velocity and heading error plots from mobile trajectory processing.
Mobile GNSS/IMU trajectory quality
Height offsets at independent control locations used to assess absolute vertical agreement of the UAV point cloud.
UAV point-cloud absolute height accuracy

Aerial imagery and LiDAR coverage were checked for completeness, UAV missions maintained RTK status, and mobile routes covered all streets in the area of interest. GCP comparisons report Easting standard deviation 0.019 m, Northing 0.023 m and height 0.021 m. Classified data were reviewed automatically and manually, while LoD2 building vectors underwent manual geometry review and Terrasolid topology checks; the final report states the building vectors were topologically consistent and complete.

Measured outputs

A controlled 3D foundation delivered in reusable geospatial formats

717 buildingsCityGML LoD2 buildings
6 blocksAerial acquisition blocks
38 missionsUAV flight missions
5 missionsMobile mapping missions
16 blocksPoint-cloud delivery blocks
0.02 m3d positioning standard deviation
0.023 mNorthing comparison standard deviation
0.021 mHeight comparison standard deviation
What this project proves

MapSoft can integrate multiple 3D capture systems and carry the data all the way to structured city geometry

CIRC-BOOST is a compact proof of end-to-end geomatics capability in a difficult urban setting: mission design, UAV LiDAR, mobile mapping, GNSS control, trajectory processing, point-cloud alignment and classification, QA/QC and CityGML LoD2 production.

38 UAV + 5 mobile missions16 classified point-cloud delivery blocks717 CityGML LoD2 buildingsCentimetre-level GCP comparison statistics

Related projects

Web map showing roof facets and selected-roof monthly/annual solar-potential values.integrated

Herceg Novi Geospatial Data & Solar Potential Geoportal

A two-phase municipal programme connecting city-scale airborne LiDAR and four-band aerial imagery, structured roof/building solar-potential data and real browser-based 2D/3D Geoportal interfaces.

3484 haFinal survey footprint

View project →
Oblique aerial view of the historic park core, designed landscape and surrounding tree cover in Bukovička Banja.integrated

Bukovička Banja Park GIS & Digital Twin Foundation

A focused protected-park project combining UAV reality capture, a 2 cm orthophoto, migration of historical CAD/Excel/vegetation information into a structured SQLite spatial database and an installed QGIS environment for practical park-data management.

17.9 haProject area

View project →
High-resolution vertical aerial view of a landscaped Novi Sad roundabout and surrounding urban streets.integrated

Novi Sad Green Areas GIS

A city-wide geospatial and software platform combining high-resolution airborne acquisition, vegetation and point-cloud processing, field-verified inventories, a spatial database, Web GIS and mobile operational workflows.

160 km²Vegetation processing area

View project →

Need a reliable 3D spatial foundation for an urban or infrastructure workflow?

MapSoft combines airborne, UAV, mobile and terrestrial capture with rigorous point-cloud processing and structured 3D data production.

Discuss your project