一种混合曲面散乱点云曲率计算方法

Curvature estimation algorithm for scattered point cloud data

针对局部点云曲面拟合过程中系数矩阵容易出现病态等问题,本文利用局部点云与切平面的夹角特性,对曲面拟合方程的权阵进行修正,并在此基础上,提出了基于三次曲面和四次曲面拟合的混合曲面散乱点云曲率计算方法,同时通过统计分析圆环面和球面的曲率计算值和真值差别,证实了该算法不仅精度高,且错误率低。

In the paper, to solve the problem of ill-conditioned coefficient matrix in fitting local point cloud curve surface, weight matrix of the blending surface fitting equation was repaired using angle characteristics between local scattered point cloud and the tangent plane. Based on the repaired weight matrix, a new curvature estimation algorithm for scattered point cloud was proposed. Statistical comparison between true value and the estimated curvature for both torus and spherical surface showed that this algorithm would be more accurate with lower error rate compared with traditional methods.