基于Hausdorff距离的点云分片精简算法  

Point cloud simplification algorithm based on Hausdorff distance and segmentation

作　　者：杨秋翔[1] 杨小青[1] 杜健[1] 

机构地区：[1]中北大学计算机与控制工程学院,山西太原030051

出　　处：《计算机工程与设计》2016年第8期2105-2109,共5页Computer Engineering and Design

基　　金：总装预研基金项目

摘　　要：提出一种基于Hausdorff距离的点云分片精简算法。对点云模型进行X-Y边界获取,保留形状特征;计算除边界外其余数据点对应的曲率,根据平均曲率构造点云模型的模糊集,引入模糊集合理论得到曲率分片的最佳阈值;以曲率最大、最小值差值的1%作为点云的曲率间隔,依次对点云数据进行分片,使某一数据点对当前分片点云的归属程度达到最大;计算各分片内数据点主曲率的Hausdorff距离,获取并保留特征点。实验结果表明,该算法提高了点云精简率,有效缩短了精简时间,较好保留了点云细节特征。A point cloud simplification algorithm based on Hausdorff distance and segmentation was proposed.X-Y boundary of all point cloud model was extracted to retain the appearance characteristics.The curvature of the remaining data points except for boundary points was calculated and the fuzzy sets of point cloud model were constructed according to the average curvature.The fuzzy set theory was introduced to get the optimal threshold of curvature segmentation.The point cloud data were fragmented one by one using curvature maximum and minimum difference between 1% as curvature interval,so that a data point was best assigned to the current fragmentation point cloud.The Hausdorff distance of principal curvature in each fragmentation was calculated and the feature points were preserved.Experimental results indicate that the algorithm improves the streamlining rate,effectively shortens the streamlining time and better reserves the detail features of point clouds.

关 键 词：HAUSDORFF距离 点云分片 数据精简 模糊集合理论 特征点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]