高级检索

几何引擎中基于光顺渐近迭代逼近的曲面覆盖

Surface Covering by Fairing Progressive Iterative Approximation in Geometry Engine

  • 摘要: 在CAD几何引擎中, 覆盖是广泛应用的关键功能, 它用一张曲面拟合给定的一条或多条边界曲线. 然而由于缺乏足够的几何信息, 很难生成满足要求的覆盖曲面, 也很少有专门解决这类问题的相关工作. 为此, 提出用光顺渐近迭代逼近(Fairing-PIA)生成单张B样条曲面覆盖给定边界曲线的方法. 与传统的光顺方法只优化单一权重不同, Fairing-PIA对每个控制点指定不同的光顺权重, 极大地提高了方法的灵活度. 首先提出一种鲁棒的方法生成拟合给定边界曲线的初始曲面, 如果边界曲线不超过4条, 使用Coons曲面作为初始曲面; 如果边界曲线超过4条, 使用点云参数化方法生成初始曲面. 然后使用Fairing-PIA提高曲面质量, 得到最终的理想曲面. 基于不同类型边界样例与传统方法对比的数值实验表明所提方法能够拟合多种类型边界, 可以通过光顺权重调节控制曲面形态, 且相较于直接求解矩阵方程的传统方法得到约2倍的性能提升, 兼具灵活性和高效性.

     

    Abstract: Covering is a widely employed key function in CAD geometry engine, which fits a surface through one or more given boundary curves. However, it is a difficult problem to generate a desirable covering surface due to the lack of enough geometric information, and there are very few relevant works that address the prob-lem. In this paper, we develop a method of generating one B-spline surface to cover given boundary curves by fairing progressive iterative approximation (Fairing-PIA). Differently from traditional fairing methods which have only global smoothing weight, Fairing-PIA can set different smoothing weights for each control point, leading to great flexibility. Firstly, robust methods are developed to generate stable initial surface fit-ting the given boundary curves. Specifically, on the one hand, if there are no more than four boundary curves, an initial surface is obtained by constructing Coons surface; on the other hand, for more than four boundary curves, a trimmed surface is generated by fitting the boundary curves as the initial surface. Then, Fairing-PIA is applied to improve the surface quality, leading to a desirable covering surface. Numerical experiments comparing the proposed method with traditional methods across different types of boundary examples demonstrate that the proposed method can fit various types of boundaries, allows for surface shape control through fairing weight adjustments, and achieves approximately twice the performance of traditional methods that directly solve matrix equations, offering both flexibility and efficiency.

     

/

返回文章
返回