Big data surface approximation from TLS point clouds

Big data surface approximation from TLS point clouds

Betreuung:  Gaël Kermarrec
Jahr:  2021


Terrestrial laser scanners generate millions of points that have to be processed efficiently for visualisation or deformation analysis. The relatively new domain of surface fitting in geodesy is gaining a high interest for many companies to meet the needs of their customers. This masterthesis deals to adress the problematic of fitting big data with iterative refinement algorithmus based on T-splines. You will implement new strategies to segment huge point clouds and ensure the continuity of the approximation. Moreover, you will be responsible to implement new development to speed the fitting, eventually without computational demanding iterations. We will test the results on diverse point clouds, from a bridge to bathimetric dataset. You will compare with other methods based on the similar principle. More specifically, you will gain experience :

  • in programming with matlab, eventually on C++ is wanted
  • on big data and its associated problematic
  • on the highly relevant topics of surface fitting
  • on answering the needs of customers

There is no need to be very good in mathematics to underestand the surface fitting, which is very intuitive. We provide some Matlab functions to be improved.

You need a great sense of creativity and humor.