The Free-form Curve And Surface Reconstruction Based On Deformable Models

To set up the effectively geometric CAD/CAM models is a hot point of the research in reverse engineering. The free-form curve and surface reconstruction, which bases on the modern measurement technology and the computer technology, has become one of the most important directions and technologies in the geometric CAD/CAM modeling. Now, there are two issues to be dealt with in the surface reconstruction. The first one is how to improve the velocity of the surface reconstruction. The second one is how to make the result of the surface reconstruction reflect the exact shape of the object accurately enough. After the deformable model was put forward, because it can fit to/ adapt to different shape with kinds of complicated and individual character of structure, and also can overcome the limitation on some special images in the traditional image processing, deformable model was concerned by researchers. And B-spline is the most popular method to represent free-form curve and surface.In this dissertation, basing on the natural science fundation of Guangdong Province project "the research on the theory and method of the reverse engineering based on vision information and CMM integration'(NO.04105386), we study the free-form curve and surface reconstruction based on deformable models and B-spline. After the systematically analysis of the theory and the research on deformable models, a new B-spline deformable model based on the finite element method (FEM) is proposed. The main work and conclusion in this dissertation are as follows: l.A B-spline active contour model based on adaptive FEM is presented. In the application of the image processing and the reverse engineering, it is a key problem how to obtain the most suitable fitting curve to express the digital curve. In this dissertation, a method based on B-spline active contour model to fit the digital curve is presented, which avoids the sampling and parameterization problems result from traditional interpolation and approximation methods and improves fitting accuracy.