Study On The Technology Of Quadric Surface And Simple Free-Formed Surface Extracting In Reverse Engineering

As an important supporting technology, reverse engineering can absorb advanced technology and accelerate the product manufacturing, so it has been paid attention by more and more manufacturing industry. The reconstruction of geometric model, which include the designing intention, is not only accurate, but also convenient for the redesign of a product. It is composed of surface extracting and constrained fitting.This thesis studies surface extracting and especially focus on the estimating of the normal vector and curvature based on scattered points clouds, extracting of quadric surface, translational surface and rotational surface. Main results achieved in this thesis are summarized as follows:1. An improved algorithm is presented to estimate the normal vector and curvature from scattered points clouds. To avoid the shortcoming of the standard algorithm, we estimate the normal vector with the standard algorithm initially, and then, the normal and curvature can be solved accurately by a iterative process, the initial normal being used as a starting value.2. Quadric surface extracting from scattered points is studied. Firstly, surface recognition is performed. Then, Genetic algorithm based real-coded and Least-Squares is used respectively to extract quadric surface from scattered points, and the merit and shortcoming of these algorithms are analyzed according to the experiment results.3.Translational and rotational surfaces extracting are studied. We use the RANSAC algorithm to estimate the translational vector and the rotational axis, because it's more accurate than the standard algorithm. After acquiring profile curve data by projection-based slicing, projection data sets are segmented based on feature points. A algorithm, based on curvature difference, is presented to acquire the tangency points. Finally, the profile curve is reconstructed by each data segment based on the segments types recognition.In the last part of this dissertation, a prototype system is designed and implemented by MFC+OpenGL under Visual C++6.0, which is a professional integrated programming entironment, to demonstrate the algorithms proposed.