Terrestrial Laser Scanning activities

The 3D-VEROS project aims at using High Precision Terrestrial Laser Scanner to improve our understanding of the present-day dynamics of various geomorphological environments. This project started in 2009 when the OSUR acquired a Leica Scanstation 2. Since then we have been building databases of high resolution surveys aiming at capturing the impact of hydrological events of various magnitude. The project has three components :

• Development of algorithms for automatic 3D post-processing of large point clouds (collaboration with N . Brodu). These include the classification of point clouds (e.g., vegetation removal), the comparison of point cloud in 3D for surface change measurement and the automatic analysis of 3D shapes (e.g., vegetation, boulders....). All these aspects are critical to use TLS data to their full extent.
• The study of bedrock river geometry and dynamics with examples from actively eroding rivers in New-Zealand. Scientific questions adressed include the measurement of bed and bank erosion rates, the analysis and evolution of roughness elements, the impact of rockfalls and landslides on river dynamics and the impact of large floods on channel dynamics. This project has been funded by INSU/RELIEF, INSU/BLANC, and is currently funded by a Marie-Curie Fellowship to D. Lague (Project €ROSNZ).

• Various applications of TLS on deformation measurement of plaster plates submitted to high temperature (with INSA Rennes), digitization of a bedrock channel running experiment, analysis of coastal cliff roughness patterns (M2 project Ludovic Guyonvarch), analysis of tidal meander development (M2 project Etienne Peyras)...

The philosophy we develop around our research on algorithm and softwares is to work with the native format of TLS (and other 3D imaging devices) : 3D point clouds (no mesh, no dems). This corresponds to our special needs given that most of the settings we are studying are characterized by 3D geometries (gorge, banks, boulders,...) with complex characteristics (roughness elements at all scales, vegetation...). We also aim at extremely precise point cloud comparison in 3D which requires a robust handling of the error budget. For this we work as long as possible with the original data wiithout any kind of projection or meshing. As articles presenting the various methods we develop will be submitted, our open source softwares will be made available with a user guide. Algorithms are developped by Nicolas Brodu and Dimitri Lague, and the coding in C++ is done by Nicolas Brodu. Enjoy them !

• CANUPO : Classification software based on Multiscale Dimensionality from Brodu and Lague, 2012 This classification software is very efficient to detect and remove vegetation in 3D, but is also a very good generic classifier of 3D point clouds according to their geometry (i.e. rough surfaces vs flat surfaces, etc...). Grab the latest sources and executables (for windows) at Nicolas Brodu website, and go to the Canupo page HERE to grab the user guide, test data and check the (hopefully) rapidly growing list of classifiers added by users.

• M3C2 : Accurate 3D point cloud comparison : first release without tutorial. The tutorial should come soon now. You can download the current version of the software and some test data of cliff collapse used in the paper. You can also find other version (linux, windows 64) and the sources on Nicolas Brodu’s website
  

Some of our datasets will be freely available at some point. But this requires a specific web infrastructure that is not yet ready. For the moment, I can only offer small excerpts, mainly used in the context of point cloud classification :

• Steep mountain channel example (Otira gorge, New-Zealand) 1.1 million points, 1cm minimum point spacing, single station scan with a Leica Scanstation 2. Also included a core point file (subsampled version with 10 cm point spacing). Ascii format (X,Y,Z). A very good example of a complex 3D scene

• Tidal pioneer vegetation (Mt St Michel Bay, France) 1.1 million points, 1cm minimum point spacing, single station scan with a Leica Scanstation 2. Ascii format (X,Y,Z,R,G,B). Relatively simple scene, but with abrupt changes in scan resolution and significant shadow behind bushes
  

• Cloudcompare : THE open source point cloud processing software for visualization, point cloud comparison, point cloud segmentation etc... An outstanding software (and it’s not because it’s french....) continuously updated by its creator Daniel Girardaut-Monteau. I use it everyday !
• Opentopography : a website with free US lidar data (mostly airborne) and useful softwares.
• Point Clouds Library : a rapidly expanding open source library dedicated to process 3D point clouds.
• ISPRS V6 working group
• LaserScanning.org.uk : interesting and dynamic forum on hardwares and softwares suites. Interesting if you are in the phase of buying a TLS equipment.
• Mike James software to easily georef SFM data : if you don’t know what is SFM (Strccture from Mtion), follow the links in his website !
• Autodesk 123D : because SFM it’s just great, and this app let’s you easily test what is SFM and why you should consider SFM in your panel of options before to buy/rent a TLS.

We use a Leica Scanstation 2 own by the Observatoire des Sciences de L’Univers de Rennes. We also use in the experimental lab a GOM Atos system, as well as a projection moiré system.

articles (pdf available on request)

• Lague, D., Brodu, N. and Leroux, J., Accurate 3D comparison of complex topography with terrestrial laser scanner : application to the Rangitikei canyon (N-Z), ISPRS journal of Photogrammmetry and Remote Sensing, 80, p. 10-26, 2013. Arxiv open access version of the manuscript

• Brodu, N. and Lague, D., 3D Terrestrial LiDAR data classification of complex natural scenes using a multi-scale dimensionality criterion : applications in geomorphology, ISPRS journal of Photogrammmetry and Remote Sensing, 68, p. 121-134, 2012. Arxiv open access version of the manuscript

conferences presentations

• Lague, D., Brodu, and Leroux, J., Unlocking the scientific potential of complex 3D point cloud dataset : new classification and 3D point cloud comparison methods, invited talk, AGU, San Francisco, 3-7 dec 2012.

• Lague, D., Bonnet, S., Davies, T. and Davy, P., Floods, landslides and bedrock river meandering, poster, AGU, San Francisco, 3-7 dec 2012.

• Leroux, J. and Lague, D., Modelling rates of bank erosion in sinuous tidal channel derived from event-based terrestrial lidar surveys in the Mont Saint Michel Bay, France, poster, AGU, San Francisco, 3-7 dec 2012.

• Lague, D., Bonnet, S., Brodu, N., Davies, T. and Leroux, J., High precision terrestrial laser scanning : measuring the individual impact of floods and landslides on meandering bedrock river dynamics (Rangitikei river, New-Zealand), EGU, Vienna, 23-27 avril, 2012.

• Brodu, N. and Lague, D., 3D Terrestrial LiDAR data classification of complex natural scenes using a multi-scale dimensionality criterion : applications in geomorphology, EGU, Vienna, 23-27 avril, 2012.

• Leroux, J., Lague, D. and Brodu, N., Spatio-temporal variability of channel bank erosion in mega-tidal salt-marshes (Mont-Saint-Michel bay) quantified by terrestrial laser-scanner surveys, EGU, Vienna, 23-27 avril, 2012.

• CPER 2008 (etat, région, conseil général, Rennes Métropole, CNRS) : achat Leica Scanstation 2
• Projet INSU/BLANC "3D-VEROS : Erosion rates measurements with terrestrial laser scanner" (PI. D. Lague)
• RTR RISC-E (Université Européenne de Bretagne) (post-doc funding to N. Brodu)
• Bourse Marie-Curie MCIOF FP7 to D. Lague : €ROSNZ : landslides, floods and erosion : new insights from event-based field measurements in the Southern Alps of New Zealand and stochastic 2D numerical modeling of long-term landscape evolution.
• Bourse de thèse CNRS (Jérôme Leroux)
• Soutien OSUR 2010 (couplages mesures TLS et instrumentation hydro-sédimentaire