多波束点云中复杂河道断面地形的自动提取方法

Automatic extraction method for complex river cross-section terrain in multi-beam point clouds

针对水下地形复杂、多波束点云数据量冗余、无拓扑结构、密度分布不均匀,高效率、高精度提取河道断面难度大的问题,提出了一种基于改进DP算法的多波束点云中复杂河道断面地形的自动提取方法。该方法根据河道场景内点云的空间分布特征剔除噪点;通过桩点将点云分割成线性数据结构体,并转换到独立坐标系内,实现点云断面提取与组织管理;分析断面曲线与拟合曲线的亲和度,保留形态特征点和平滑点、剔除冗余点,完成断面形态精化。定性结果表明:该方法自动化程度高、形态特征点保留完整、冗余度有效降低。定量评价表明:面积差百分比平均值为0.02%、冗余度降低百分比平均值为80.85%,在降低数据冗余度的同时,河道断面形态、面积精度未受损失。

A method for automatically extracting cross-section terrain from complex underwater terrain,and redundant multi-beam point cloud data,without a topological structure and uneven density distribution,has been proposed based on an improved Douglas-Peucker algorithm.The method eliminated noise points based on the spatial distribution of the point clouds in the river scene.Linear data structures were created by segmenting the point cloud into sections using pile points,which were then transformed into an independent coordinate system to facilitate the extraction of cross-sections and the organization of point clouds.The affinity between the crosssection curve and the fitting curve was analyzed to preserve the characteristic points and smooth points while removing redundant points,finally resulting in a refined river cross-section.The qualitative results showed that the method has a high degree of automation,retains the complete characteristic points,and effectively reduces redundancy.Quantitative evaluations showed that the average percentage of area difference was 0.02%,and the average percentage reduction in redundancy was 80.85%.The accuracy of the shape and the river cross-section area were not lost while reducing data redundancy.