Bozkov dolomite caves survey

Václav Štěpán, Lenka Thinová, Tomáš Bláha, Lukáš Loula, Dušan Milka, Vratislav Ouhrabka, Martin Kaschner

2024-09-12

Bozkov Dolomite Cave

Courtesy of CACR [1].

LiDAR data acquisition

LiDAR data acquisition

Zeb Horizon scanner [2]

Data were acquired in a joint effort as a technology demonstration:

• Dušan Milka and Vratislav Ouhrabka provided the cave
• Tomáš Bláha provided the scanner
• Lukáš Loula processed the scans
• Václav and Martin did mesh and AR

Raw data processing¹

• Raw data from LiDAR and position/movement sensors processed into noise filtered high resolution point clouds.

Postprocessing²

• Point clouds cleaned and decimated to 5 cm resolution.
• Converted to mesh, decimated to 100 kilofaces.
• Scale down, shifted, set up in model-viewer [3].

1. Data processing by Lukáš Loula, GeoSLAM Ltd.

2. CTU, using Agisoft Metashape and Blender

Point cloud presentation

Point cloud

Challenges

• Original dataset is 190 Mpoints, cca 2 GB of zipped data
• Limited support for mobile platforms/browsers

Solutions

• Dynamic streaming from multi-level dataset
• Conversion to mesh representation

Dynamic streaming options

Potree [4]

• Open source WebGL point cloud viewer for large datasets
• LAS data are converted to octree structure
• Viewer deployed on user server

ArcGIS

• Enterprise 3D GIS solution
• Data hosted in cloud and streamed on demand

• Static JavaScript scripts load the data on-demand
• Deployment on user premises
• Requires simple code editing to deploy
• Limited mobile support (UI not optimized)

Point cloud limitations

Limitations

• Too low information density
• Poor mobile support

Solution

• Conversion to mesh geometries

• Too much data for CAD/drawing
• Poor mobile device visualization support
• Simplification necessary

Mesh geometries

Point cloud to Mesh conversion

Step 1: Agisoft Metashape

• Data decimated from 190 Mpoints to 3 Mpoints
• Point cloud converted to TIN mesh (1.3 Mfaces)
• Mesh simplified to 100 kfaces

Step 2: Blender

• Data scaled 1:100 for presentation in AR/VR
• Shifted to origin, saved as gITF
• Converted to wireframe tube structure, saved as gITF.

• Textured meshes are used in game industry
• Mobile support is abundant
• Details stripped/limited, but
• AR/VR visualization options

Wireframe representation, 100 k faces, shown using model-viewer [3].

Dare to open this?

(Scan the QR code to open the model on mobile device)

Using augmented reality

Screen recording of iOS AR preview of the model

Summary

• Handheld SLAM LiDAR are fast
• Point clouds can be published on premises an in cloud
• Meshes are better suited for web/mobile/AR presentation

How can you make a 3D model

Photogrammetry

• Compute the scene from photos using open source software
  • Meshroom [5] for general photogrammetry
  • OpenDroneMap [6] for UAS mapping

LiDAR & Camera on the phone

• Find a friend with iPhone Pro and use the built-in LiDAR
• Try PolyCam [7], Pix4Dcatch [8] apps

Surface gamma radiometry

GT-40 spectrometer

• Portable gamma spectrometer.
• Cylindrical NaI(Tl) 3x3” crystal
• Device carried on operator back in a harness.
• Recording (1 s^-1), H*(10) [nSv/hr], GPS position/altitude.
• Capability to identify radionuclides, on-the fly assay modes.
• USB & Bluetooth & WiFi

Where is the cave in that?

Data fusion in 3D using ArcGIS

• LiDAR point clouds are GPS georeferenced using above ground points
• Data overlayed in 3D with:
  • Height map (ČÚZK LiDAR surface survey)
  • Geological map (ČGS)
  • Orthophoto map

Radon measurements

TSR4 probe locations

Radon monitoring using TSR4

Official monitoring

• RAMARN integral dosimeters, average ERC from 3 locations along visitor path
• Summer and winter seasons (April–September, October–March)
• Effective dose calculation based on hours underground, average ERC, DCF and “cave factor” (depends on unattached fraction)
• New approach with DCC instead of DCF = increase of number of seasonal workers

Continuous monitoring in the cave for research

• locations Peklo and Jezero, RADIM 3, 3A monitors
• continuous measurements since 2012 on 7 locations on visitor path and 2 probes in ventilating openings, TSR4

Seasonal averages of ERC (RAMARN)

Data courtesy of: Petr Zajíček, administration of the caves of the Czech Republic

E (mSv) for guides and students

Data courtesy of: Petr Zajíček, administration of the caves of the Czech Republic

Appendix

Did you like the presentation?

Last version is available at: https://linux.fjfi.cvut.cz/~w/240529-Bozkov

Solid/semitransparent visualization, 100 k faces, shown using model-viewer [3].

References

[1]

Bozkov dolomite caves [online]. [accessed. 2024-05-20]. Available at: https://bozkovske.caves.cz/en

[2]

ZEB-HORIZON user manual [online]. 2020 [accessed. 2024-05-25]. Available at: https://geoslam.com/wp-content/uploads/2021/02/ZEB-Horizon-User-Manual-v1.3.pdf

[3]

Model-viewer [online]. [accessed. 2024-05-20]. Available at: https://github.com/google/model-viewer

[4]

Potree 1.8.2 [online]. 2024 [accessed. 2024-05-25]. Available at: https://potree.org

[5]

AliceVision photogrammetric computer vision framework [online]. 2024 [accessed. 2024-05-25]. Available at: https://alicevision.org/

[6]

OpenDroneMap [online]. 2024 [accessed. 2024-05-25]. Available at: https://www.opendronemap.org/

[7]

Polycam - LiDAR & 3D scanner for iPhone & android [online]. 2024 [accessed. 2024-05-25]. Available at: https://poly.cam

[8]

PIX4Dcatch [online]. 2024 [accessed. 2024-05-25]. Available at: https://www.pix4d.com/product/pix4dcatch/