保留边界的点云简化方法  

Point cloud simplification with boundary points reservation

作　　者：黄文明[1] 肖朝霞[1] 温佩芝[1] 吴晓军[2] 

机构地区：[1]桂林电子科技大学计算机与控制学院,广西桂林541004 [2]哈尔滨工业大学深圳研究生院自动控制与机电工程学科部,广东深圳518055

出　　处：《计算机应用》2010年第2期348-350,384,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(50805031);广西科学基金资助项目(桂科自0991240)

摘　　要：针对点云简化算法中边界点丢失的问题,提出了一种保留边界的三维散乱点云的非均匀简化算法。首先利用kd-tree建立散乱数据点云的空间拓扑关系,计算出每个数据点的k邻域;然后针对目前依据点云分布均匀性算法提取边界效率低的问题,提出一种改进的点云边界点判定算法;最后保留所有边界点,对非边界点,根据曲面变分值和k邻域点已保留比例,进行点云的非均匀简化。实验结果表明,该算法精度高,空间复杂度低,而且简化后点云边界保留完整。A non-uniform simplification approach with boundary points reservation was proposed concerning the boundary points loss caused by most point cloud simplification algorithms. First, kd-tree was used to represent the spatial topology relationship of the scattered point cloud and to calculate the k-nearest neighbors for each data point. Then an improved algorithm for boundary points detection of point cloud was presented to solve the low efficiency when the current algorithms extract boundary according to uniform distribution of point cloud. Consequently, all the resulted boundary points were reserved, while the non-boundary points were reserved for final non-uniform simplification according to the surface variation and their nearest neighbors. The experimental result shows that the proposed simplification approach has high accuracy and low space complexity, and can reserve the boundary points.

关 键 词：边界点 非均匀简化 散乱点云 kd-树 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]