| 李伟华,刘良林,李久明,吕超.联合点曲率值与竖直平面约束的点云简化算法[J].井冈山大学自然版,2023,44(6):84-92 |
| 联合点曲率值与竖直平面约束的点云简化算法 |
| A POINT CLOUD SIMPLIFICATION ALGORITHM BASED ON COMBINING POINT CURVATURE AND VERTICAL PLANE CONSTRAINT |
| 投稿时间：2023-05-04 修订日期：2023-09-20 |
| DOI：10.3969/j.issn.1674-8085.2023.06.011 |
| 中文关键词: 点曲率值 竖直平面约束 关键点 信息熵 点云简化 |
| 英文关键词: point curvature value vertical plane constraint key point information entropy point cloud simplification |
| 基金项目:国家自然科学基金项目(42207208);江西省教育厅科技计划项目(GJJ2201631);井冈山大学自然科学基金项目(JZ2002) |
|
| 摘要点击次数: 89 |
| 全文下载次数: 128 |
| 中文摘要: |
| 针对现点云简化方法计算量通常为点云量指数级倍数、简化精度受邻域设置影响大的问题,本研究依据物体表面凹凸起伏的空间特征建立微小的竖直平面,先由第一号点为种子点建立竖直平面约束条件,检测处于竖直平面内或与其距离小于阈值的点,通过所检测的点计算其内每个点的曲率值,提取曲率值符合要求的点存储于简化点集与种子点集。下一次运算从种子点集中依次选取种子点,同理可计算出简化点与新种子点,将此时运算的种子点从种子点集中删除,依次循环,当种子点集为空时,算法结束,此时简化点集与剩余散点融合即为简化结果。通过实验与其它方法比较,本方法可将物体点云集Dino和Elephant的简化率分别从网格法的60.39%和80.148%、信息熵法的76.628%和76.072%提高到85.23%和82.901%,并缩短了运行时间。此方法在保证较高精度的同时,可大幅度去除冗余点云,为点云存储与后续算法设计带来便捷。 |
| 英文摘要: |
| Aiming at the problem that the calculation amount of current point cloud simplified methods is usually an exponential multiple of the point cloud amount, and the simplification accuracy is greatly affected by neighborhood settings, this study establishes a small vertical plane based on the spatial characteristics of the object's surface undulations, the first point establishes a vertical plane constraint condition for the seed point firstly to detect the points within the vertical plane or at a distance less than the threshold. The curvature values of each point within the detected points are calculated, the points with curvature values that meet the requirements are extracted and stored in the simplified point set and seed point set. The next operation selects seed points sequentially from the seed point set, and similarly, simplified points and new seed points can be calculated. The seed points for this operation are removed from the seed point set, and the algorithm is completed when the seed point set is empty. At this point, the simplified point set is fused with the remaining scattered points to obtain the simplified results. Compared with other methods through experiments, this method can increase the simplification rates of Dino and Elephant in object point cloud sets from 60.39% and 80.148% of the grid method, 76.628% and 76.072% of the information entropy method, to 85.23% and 82.901% respectively, and shorten the running time.This method can significantly remove redundant point clouds while ensuring high accuracy, bringing convenience to point cloud storage and subsequent algorithm design. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
