Abstract: A method of interactive extraction of boundary of specified target feature is proposed to extract the boundary of specified target feature from the scattered point cloud with multiple feature boundary. Firstly, the point cloud of specified feature and its surrounding is picked up. The topological relations of scattered point cloud are established with [KD] tree, then the k-nearest points of each point would be searched. Whether the point is the boundary point is judged by the maximum value of angle difference which is calculated by projection points on the least square plane with the k-nearest points.  Lastly, one point on the boundary of the specified feature is picked up as the seed point, then according to a fixed direction to search for boundary points, until the entire closed boundary is extracted. Experiments show that, this method can accurately
obtain any specified target feature boundary on the scattered point cloud.

Key words: scattered point cloud, boundary extraction, interactive, target feature

摘要： 为了在具有多个特征边界的散乱点云上提取指定目标特征边界，提出了一种基于目标特征边界交互提取的算法。拾取目标特征及其周围的点云。利用[KD]树建立散乱点云的空间拓扑关系，计算出每个数据点的[k]邻域。通过数据点[k]邻域构成的最小二乘平面的投影点角度差，根据角度差的最大值来判断该点是否为边界点。拾取目标特征边界上的某个点作为种子点，按照某一个固定方向搜索边界点，直到搜索整条封闭边界为止。实验表明，该方法能够准确获取散乱点云上任意指定目标特征的边界。

关键词: 散乱点云, 边界提取, 交互性, 目标特征