Abstract: A new automatic extraction algorithm of boundary characteristic point is proposed. The points in small neighborhood of the sampling point establish the tangency plane by using least square method. And these points project to the tangency plane. The boundary characteristic point is detected according to the geometric distribution characteristics of the point set in the projection plane based on the theory that the sum of the field power of every point can reflect the average function of the point set. Using the bidirectional search algorithm of the nearest point, the extracted boundary characteristic point is sorted and the boundary of scattered point is automatically created. Experimental result indicates that the algorithm can extract the boundary of point cloud quickly, accurately and effectively.

Key words: scattered point cloud, boundary extraction, neighborhood

摘要： 提出一种新的散乱点云边界特征点提取算法。根据点云数据小邻域内点用最小二乘法拟合建立微切平面，并将这些数据点向其微切平面投影，利用点集中每个点的场力大小之和可以体现点集平均作用的理论来分析投影面上点集的几何分布特性，据此检测边界特征点。利用双向最近点搜索算法对提取出来的特征点进行排序并自动生成边界曲线。实验结果证明该算法能够快速、准确、有效地提取点云的边界。

关键词: 散乱点云, 边界提取, 邻域