中文摘要：

针对传统平面参数化不能直接对零亏格封闭模型参数化,而球面参数化较为适用于形态接近于球体模型的问题,定义了一种超圆柱面的参数域表示,并提出一种渐进的超柱面参数化算法,对于形态接近于圆柱或具有单支简单骨架的网格模型能够提供较小的几何形变;基于此,提出一种骨架引导的网格模型圆柱面参数化方法,能够对具有复杂骨架结构的网格模型实现全局参数化,它通过分块参数化由骨架信息得到的网格模型主体和分支,进而将分支圆柱面参数网格配准、优化并拼接至主体参数网格得到无缝的全局参数网格;对于人体和动物等具有明显骨架结构的网格模型,骨架引导的圆柱面参数化具有较小的几何形变.最后给出了实验结果,并展示了圆柱面参数化在约束纹理贴图中的应用.

英文摘要：

Previous planar parametrization methods cannot be applied to 0-genus closed mesh models directly and spherical parametrization is mainly applicable to the models whose shapes are close to the sphere.In this paper,we define a hyper-cylindrical parametric domain and present a progressive cylindrical parametrization algorithm.Such an algorithm can be applied to the mesh models which have a simple skeleton or resemble the cylinder.We further present a skeleton-guided mesh cylindrical parametrization method such that the models with strong skeletons can be globally parameterized.The parametric domain is purely determined according to model shape,relieving geometric distortion caused by parametrization.The model is segmented into the main body and several branches.We parameterize each of them and finally stitch the parameterized branches onto main body.We verify the effectiveness of the cylindrical parametrization method through experiments and show its application to constrained texture mapping.