I have a question concerning a paper. In it the authors try to approximate a NURBS curve by biarcs and to do so, they first need to find a polygonal approximation of the NURBS curve. They chose a geometric decomposition procedure which they begin defining as follows:
The geometric decomposition is essentially a classical subdivision scheme based on the convex hull property of NURBS [6]. The process is outlined as follows.
1.Decompose the NURBS curve into piecewise Bezier segments.
Which already starts raising questions for me, how is it possible? Isn't the points of NURBS to produce curves which Bezier curves can not, like circles? Or should I assume that the Bezier curves the authors mention here are rational? And how one should perform such a decomposition? The next steps in the procedure are also not exactly clear, but I am unable to post them here due to formatting difficulties. The part of text I am discussing can be found at page 3 of the article.
Any help would be greatly appreciated! Thank you very much and have a nice evening!