The Algorithm Of The Copy Of Feature Curve Network Between Two Similar Multi-line Shaped Surfaces

Subdivision surfaces play an increasingly significant role in the three-dimensional reconstruction,when we have the curve network which meets the rules of the two-dimensional manifold, we can getcontrol meshes of complex physical, thus can reconstruct the complex body through subdivisionperfectly. Based on the latest study of subdivision surface and features reuse of curve network, drawlessons from features reuse method of pattern, we focus on features reuse algorithms of curve network,including the fairing algorithm of multi-line shape, three-dimensional deformation algorithm based onnearest point iterative algorithm, deformation algorithm based on shrink wrapping approach anddistortion correction algorithm of curve network, the main research results are as follows:1. The Gauss smoothing algorithm of multi-line shape is proposed, the planar discretecurve smoothing algorithm is applied to the multi-line shape, we can get the smooth curve onthe different scales space, using three tangent vector equation based on the three orderspline interpolation and Ferguson curve to solve the normal and curvature, the normal of plane curvelays the foundation for the evolution of the curve, curvature is used to judge the corner point.2. Propose the method of smooth-deformation-mapping-amend to solve the problem of featurecopy, deformation is divided into initial deformation and registration deformation, initialdeformation algorithm can also be divided into two methods: the first one is a hip femoral example,according to their own individual characteristics of the femur, using the characteristic line, featurepoint alignment methods to achieve the deformation of multi-line shape, the other one is inspiredfrom the iterative closest point algorithm of point cloud registration (ICP: Iterative ClosestPoint), This algorithm is simple and intuitive, easy to implement, can establish a good initialposition before deformation. For the registration deformation, through the shrink wrapping algorithm,we can realize the curve evolution of plane curve, the normal of discrete curve as theattractive direction, the curve between two discrete spacing of1(in image it represents two adjacentpixels) as the relax direction, with the common role of attractive and relax, the evolution of planecurve can realize through several iterations, we can establish the mapping relationship between twocurves by parameterization method based on curve.3. Set up the hierarchical mapping between two similar multi-line shapes, the problem ofdeformation transform into a planar curve evolution problem. After established mapping function by deformation, we can get the feature curve network after mapping. But because there may existerrors after mapping, we put forward the wrong point correction method and curvature correctionmethod of feature curve network.4. With the image processing software DateSet and Matlab, we successful implementation ofthe deformation of multi-line shape of the initial deformation and registration deformation, establishthe mapping function of feature curve network, and modified to establish the feature curve networkmeet the rules of two-dimensional. Numerical experiments show that, the algorithm proposed in thisartical is correct and feasible, have wide application, and can quickly, accurately realize the copy offeature curve network.