Improving NURBS Surface Sharp Feature Representation
Nallig Eduardo Leal, Oscar Ortega Lobo, John William BranchAbstract: —Surface fitting is one of the most important stage of the 3D reconstruction process, since in this stage the computational model of the real object is obtained. NURBS is a surface fitting method widely used that has become a standard in CAD/CAM system due to its stability, flexibility and local modification properties. Despite of the advantages of fitting with NURBS, it is still necessary improve the representation of sharp features like high curvatures, edges and corners with this fitting method. Find a correct parameterization of the NURBS can contribute to improve the representation of sharp features, even though the manipulation of the NURBS parameters imply deal with non linear problems when fitting. In this paper, a new method for improving NURBS surface sharp feature representation is presented. The method first subdivides the fitting data in clusters, by using SOM. Then, in each cluster uses an evolutionary strategy to obtain the weights of the NURBS so that the error fitting is minimized and the representation of sharp features is improved.
Keywords:
—Evolutionary Strategies, NURBS, SOM, Surface optimization.
doi:10.5019/j.ijcir.2004.97
Full Text PDF
^ TOP