Research On Feature Line And Surface Extraction In Reverse Engineering

Surface reconstruction based on coordinate measuring is widely used in such areas as reverse engineering, computer vision and reconstruction of medical image. Feature extraction of noisy arbitrary triangle mesh, as one of the critical and difficult problems in the complicated surface reconstruction, is researched in this thesis.Firstly, the dissertation presents a feature line extraction method of Geometric second consecutive: the feature points were extract from triangle mesh or original point cloud, then they were connected and organized into "rough" feature lines. Then in order to meet the need for practical industrial application, an optimal and recover algorithm is introduced to smooth the noisy feature line by branch cutting, edge collapse, edge split and edge swap.The dissertation studied the methods of data, pre-process and presented the algorithm of 3D mesh segmentation: Using curvature, vector and the prerequisite feature lines to detect the segment edges, then a region growing method is applied based on those edges is applied to segment the triangle meshes.At last the dissertation presented the algorithm of feature surface recognition based on quadric surface fitting and geometry property. Firstly using quadric surface to fit the given surface and calculate the error metric to identify free surface and general quadric surface. Secondly computing geometry property of each point on Gaussian Sphere and Curvature Coordinate, The distribution will identify each kind of quadric surface and blend surface. This algorithm is easy to recognize plane, cylinder, cone , spheres, blend surface and free-from surface.