Research On Modeling And Application Of Reverse Engineering Surface Reconstruction

Current reverse engineering (RE) consists of some stages such as digitization, pre-treatment of data cloud, segmentation, surface reconstruction (SR) and so on. Keys of each stage are still the hot subjects in this discipline. SR is one of the most important, whose speed and quality eventually constrain the downstream procedures, for example, CAD design. Currently in RE based physical objects, according to the difference of RE CAD system for modeling, methodologies for SR can be categorized into two types: Classical surfacing ( CS) and Rapid Surfacing ( RS ) . CS abides by the classical RE process flow, i.e., from point to curve to surface. It uses Bezier or NURBS surface to construct target surface with feature boundaries or point clouds. While RS does the same thing just through the polygonization of data sets. A complete polygonization includes: firstly create triangulation of point clouds, and then segment it into quadrilaterial networks, and then parametrization the polygon framework, finally to fit every framework with NURBS to get a surface CAD model with a certain patch continuity for follow-up treatment.Advanced surface management is performed to analyzed the aesthetic and engineering design implications and to develop high-quality surfaces. Unfortunately great time investments and skilled operators are required, especially if class A surfaces are needed. RS has been introduced in order to overcome the above disadvantages. However, RS techniques introduce some drawbacks too. The capability of quickly adapts surfaces to the point cloud requires the use of higher degree NURBS surfaces with respect to CS that uses the lower. Moreover, curvature continuity between patches is often difficulty to obtain, resulting in several manufacturing limitations.As a result, firstly the thesis explores the basic theories and applications in modeling of the above two surfacings, and develop a new methodology——hybrid surfacing (HS) which integrates the surfacing advantages of the former two. As for HS, use RS to polygonize the point cloud to extract the patch templates with quadrilaterial parametrization types, then take further treatment in CS system and construct surface model. It can not only efficiently overcome the drawback of curvature continuity in RS, but also reduce the difficulty of feature segmentation in CS, resulting in a rapid surface reconstruction.