Feature Recognition And Surface Reconstruction In Mesh Model

Reverse Engineering (RE), as a method of product designing and manufacturing, has been paid more and more attentions by the whole world during last 90's.With the development of computer science and measure technique, it has already become as a key way of new product designing and advanced technique absorbing.Surface reconstruction, aiming to achieve the mathematical expression of geometric model for complex free-form surface from the scattered data, is a key technique of Reverse Engineering. Major commercial CAD software takes Non-Uniform Rational B-Spline (NURBS) as main expression presently.Through studying on the surface reconstruction upon randomly sampled points, improved surface fitting methods are put forward in this paper, aiming at triangle mesh. We explain the ways of surface reconstruction in details, including quadric surface, extruded surface, revolved surface and free-form surface.Least-square method is widely used to generate quadric surface at present, but the efficiency of it is low because of its non-collinear. With the help of Gaussian image, we propose a method here that denoise the mesh model first, then extract triangles which make up of the surface feature, finally we reconstruct the demanded surface. The exam results show that this approach is available and efficient.Extrude surface and revolved surface are also common in Reverse Engineering. The key of algorithm is to extract extruding direction or revolve axis direction. Based on deep analyzing on Gaussian image, we get the extruding direction magically. Also, we gain the revolve axis direction according to its primary directions.Free-form surface fitting is a hotspot in RE. The results of constructing surface directly upon randomly sampled points are always unacceptable. In this paper, we try to form ordered(m*n) points according to the mesh model, then we fit the given ordered points. Experiments show that the results are much better.