Surface Modeling And Precision Evaluation For Wrap-around Models Based On Scattered Points

With the rapid development of the current digital measuring technology, Reverse Engineering has been widely applied in image processing, computer vision, geometric modeling and digitalization manufacturing etc. Surface reconstruction is one of the key technologies of reverse engineering. The wrap-around model reconstruction from dense scattered points (called "point cloud") is very important to shape design and manufacture of ship, automobile and aeroplane. It is the keystone and nodus to shape a high precision smooth surface. It is difficult for scattered points only with coordinate information to be parameterized equably, rationally and availably. Currently, surface reconstruction is implemented mostly interactively. In this paper, taking the scattered points obtained from laser scanner for example, based on Non-uniform Rational B-splines (NURBS), the extraction of boundary points from scattered data points, parameterization, model reconstruction and precision evaluation are systemically studied, a new automatic model reconstruction method is proposed. The main research contents and achievement are as follows:Firstly, with consideration of the difficulty of surface reconstruction from dense scattered points, an automatic tracing algorithm of the wrap-around model boundary and an automatic data parameterization method are presented. On the basis of these, the sequential data set is obtained by means of local tangent space parameterization, and the data noise is eliminated effectively. At last, the surface reconstruction is finished by means of skinning method. This algorithm avoids disadvantage of interactive means during the existing modeling process. The result demonstrates that the proposed algorithm is efficient, the fitting accuracy and the running time are improved obviously.Secondly, taking the reconstructed CAD model for example, the unconstrained matching is discussed. The surfaces are matched roughly at five-key-point positions including center point and four corner points which can speed up the matching. And SVD method is used for accurately match to improve the matching accuracy. The distance of the sample points to the CAD surface is calculated for evaluating precision.Finally, in the integrated development environment of VC++6.0 with the OpenGL three-dimension graph standard library, the technologies and key algorithms presented above are implemented in automatic modeling and locating system for the large thin-wall part. One detailed example shows the feasibility and application value of the system.