基于截面特征相容的曲面约束重构

Study on feature compatibility in surface reconstruction

为了从点云中重建高品质的曲面模型,提出一种基于截面特征相容的曲面重构方法.该方法将由点云切片获取的截面曲线分层次表达为直线、圆弧、B样条等基本曲线元和特征点的集合,并基于曲线微分几何属性,建立截面曲线特征点对应的数学模型,采用动态规划法求解并确定其初始对应关系,以曲线元特征匹配准则优化对应,逐步确立截面曲线间的特征对应关系,将特征对应关系映射为参数对应即可利用成熟算法逼近生成蒙皮曲面.实例表明,该算法获取的曲面模型不但满足精度和光顺性要求,而且忠实地反映了原始物理样件所蕴含的特征信息.

A surface reconstruction method based on feature compatibility of sectional curves was proposed to reconstruct fair and shape-preserved surface from point cloud. This method hierarchically described each section curve as a set of curve primitives, i.e. line, arc and B-spline curve, and connecting points. According to the differential properties of curves, a scheme that closely approximated the optimal relative matching between the connecting points of the curves was constructed. An initial corresponding relationship among the sectional curves was established through solving the equation by the dynamic programming method. Then, the relationship was optimized by matching the curve primitives and making the knot vector of each sectional curve be the same. A fair surface was generated by lofting the sectional curves. Applications showed that the proposed algorithm is feasible.