Abstract: A method of [C1] surface reconstruction based on cross-sectional data is proposed in this paper. Blending method is used to merge the points and curves in order to generate a smooth reconstructed surface. The proposed method is capable of surface fitting directly based on the cross-sections where the number of control points varies from section to section, which avoids the step of having compatible cross-sectional curves defined on common knot vector in NUBRS surface reconstruction. So it won’t end up with a large number of control points in the resultant skinning surface due to knot inserting. Besides, the method also completes the curve and surface merging by using blending method, which makes the differential information not necessary in smoothing joining.

Key words: surface reconstruction, cross-sectional data, spline, blending method, surface joining

摘要： 为了避免NURBS曲面重建需要进行节点矢量相容的问题，提出了一种双方向融合插值的[C1]参数曲面重建方法，该方法先后分段插值截面上连续的数据点、截面曲线以构造样条曲线和曲面片，并引入融合算法进行曲线、曲面拼接，从而得到光滑的待建曲面。该方法不会产生由节点插入所带来的大量的数据冗余以及复杂的计算过程，同时采用了融合的思想来处理曲线、曲面的拼接，改良了传统参数曲线、曲面拼接方法需要满足边界条件的缺陷。

关键词: 曲面重建, 截面数据, 样条, 融合算法, 曲面拼接