Approximate a Surface

Hi all,
maybe someone has the same problem and got a solution.

I have a list of unsorted points. I can Interpolate them with GeomAPI_PointsToBSpline. That works fine. But if i try to interpolate them to a surface with GeomAPI_PointsToBSplineSurface, i didnt get successful results, because the points are not sorted. How can i get TColgp_Array2OfPnt from TColgp_Array1OfPnt ready for the interpolation. Or is there an other function for interpolation.

thanks, juri

Francois Lauzon's picture

Hello Juri,
you might want to look at GeomPlate_BuildPlateSurface, you can build a NURBS surface from unsorted points. One of the sample (...\samples\standard\mfc\11_Plate) show you how to use it.

Good Luck,
Francois.

juri's picture

Hello Francois,

thank you for the hint. The code from the sample works fine and was a big help.

bye, juri

LI's picture

This method works fairly well. But an application condition must be pointed out: (In inverse steps)

Suppose that some points are picked from a surface. We use these points to reconstruct a surface by the plate method. It's easy to see that if the projection of the original surface on the plane XOY has self-intersection, the surface construction from these points may fail.
Under this constance, the plane XOY should be a local plane on which the projection of the original surface has no intersection.
Since the original surface doesn't exist, there is still some work to do to insure the construction.

Abraham4D's picture

I am wondering what is the parameters in this example "11_Plate" that help us control the amount of approximation (e.g. If I want to keep the approximation as less as possible like least square method) Any help please please please.

LI's picture

Could you please send me the points? I'd like to have a try.

To: yuming.li@univ-reims.fr

Thanks!

Abraham4D's picture

I am wondering what is the parameters in this example "11_Plate" that help us control the amount of approximation (e.g. If I want to keep the approximation as less as possible like least square method) Any help please please please.