Modern measuring systems can easily produce data sets consisting of thousands, even millions of points.
The data points are almost always random.
Also, accurate modelling of shapes using NURBS often involves surfaces with tens of thousands of control points.
The developed surface approximation method was designed to address such situations. Using the cloud of points and a suitable base surface as input, the process involves the following iterative steps:
- Data parameterisation (assigning UV coordinates to data points)
- Linear Least Squares fitting
- Knot insertion
Knot insertion is usually necessary in order to introduces additional control points and to provide additional flexibility for modelling the surface detail. The iterative process may be applied a number of times until satisfactory modelling accuracy is achieved.
The described process is known to suffer from two main issues, which have been successfully resolved.
NURBS
Contact us
Dr. Mihailo Ristic
Senior Lecturer
T: +44 (0)207 594 7048
E: m.ristic@imperial.ac.uk
Dr. Djorje Brujic
Research Fellow
T: +44 (0)207 594 7175
E: d.brujic@imperial.ac.uk