Under development since 2004 by EDF's R&D, CloudCompare processes and compares dense clouds of 3D points. It is used mainly to process internal generating plant data using topographic laser scanners.
- calculation of local distances between two dense clouds of 3D points (left image)
- filtration of laser scanner measurement noise to reveal true differences (centre image)
- segmentation of differences to reveal objects or fragments of objects corresponding to the differences between the two datasets compared (right image)
A UNIQUE SOFTWARE PACKAGE WITH MANY STRENGTHS
There are several unique aspects to CloudCompare:
- the data structures used: an "octree" enables large point clouds (of several million 3D points) to be loaded into memory and displayed, and the differences between 2 large datasets to be calculated in just a few seconds. A Kd-tree' is also used for high-speed re-matching of two 3D point clouds
- the 2 types of remote calculation between 2 point clouds, which did not exist before: precise calculation based on the Hausdorff distance method, and high-speed (less accurate) calculation based on chamfer distance; the distances can now be assigned symbols
- the filtration of measurement noise and the many scalar field manipulation tools
- the ability to take account of sampling differences between the compared datasets
- the ability to take account of scanner visibility for each dataset
- the EDL (Eye Dome Lighting) rendering of point clouds via the graphics card, which significantly improves the on-screen legibility of this type of data
- the direct inclusion of multiple libraries (PCL, CANUPO) in the software and the ease with which new plug-ins can be created
The major strengths of CloudCompare are its speed of execution, and the absence of freeware competition.
PRACTICAL EXAMPLES OF CLOUDCOMPARE IN USE
CloudCompare enables UTO to measure the variances between laser measurements made inside a power generating plant and the CAD models reconstructed from these measurements in order to detect errors or oversights.
The LNHE (National Hydraulic and Environmental Laboratory) uses it upstream of TELEMAC and in response to the need for hydraulic research studies. CloudCompare now enables the LNHE to:
- conveniently view clouds containing tens of millions of data points
- filter and sub-sample a point cloud on the basis of a broad range of criteria (local curvature or altimetry gradient in particular)
- compare two point clouds by calculating local (distance) variances
- create a mesh from a point cloud or rebuild a CAD model from a point cloud
- lcally reduce or increase the level of mesh detail on the basis of a broad range of criteria (local curvature or altimetry gradient in particular)
This software has been under development since 2004 in the context of a CIFRE thesis coordinated jointly by EDF's R&D and Télécom Paris.