Indoor 3D modeling

Multinnov has developed UAView Optical Flow capable of high-resolution photogrammetry. What are the interests in being able to do 3D modeling?

It is simply possible to locate defects in space without trying to sort through 1000 photos that are difficult to locate indoors using a plan (for example). Locating, quantifying and measuring defects becomes possible with photogrammetry.

It is also possible to calculate the volume and dimension of an underground gallery, a room or any other cavity being indoors. Knowing the shape, the elevation of a pipe from the inside, the talwegs and/or water retention via the use of orthomosaics .

Our technical solution combined with photogrammetry allows us to determine the distance of a defect. This can be very important in piping, especially to possibly plan in-depth work.

For the final customer, the intervention with UAView Optical Flow will save time, reduce human risks and save money. The 3D model will provide a unique overview, allowing the right decisions to be made following the technical expertise.

It is therefore the guarantee of a complete and efficient intervention.

 

Data acquisition

Set the camera to Time Lapse mode with a time interval of your choice. Photogrammetric acquisition is relatively classical. Retrieve the photo file and compile it in your photogrammetry software according to the desired parameters. To calibrate the scale of your model, it is essential to take dimensions on site in order to nest these dimensions in the photogrammetry software.

Presentation of results in an underground pipeline 

Dimensioning a structure:

This will allow us to know storage volumes, calculate flow rates, check for potential structural deformations…

Locate a defect on the 3D model, then locate the raw image in order to possibly deepen a defect.

Raw pipe scan data

You can retrieve the raw data in order to carry out the photogrammetry of this part of the underground pipeline yourself: Upload link

Settings of the above modeling:

  • Software: Metashape
  • Alignment: 465/493 alignées
  • Points cloud: 342,828 points (average precision with generic method)
  • Dense points cloud: 17,512,622 points (medium quality and light depth filtering)
  • Mesh: 1,048,873 faces (meshing via dense point cloud, active interpolation, medium quality)
  • Texture: 4096 x 4096 (generic mapping and mosaic fusion)