移动最小二乘重采样在三维重建中的应用

Application of mobile least squares resampling in 3D reconstruction

针对地面三维激光扫描仪采集的多站点点云数据受配准精度和重叠区域的影响,经多视角对齐后生成带有大量噪声和冗余的散乱点云从而影响曲面重建的问题,提出一种基于移动最小二乘重采样的算法:计算点云的k邻域,建立散乱点云的空间拓扑关系,选择合适的基函数和权函数,建立局部拟合区域的拟合函数,结合体素化网格模型实现点云的重采样,分别使用Crust算法和逆向工程软件Geomagic Studio对重采样的点云进行曲面重建。结果表明:该算法在保证局部细节特征清晰的基础上,能够提高模型表面的光滑性和三维重建的效率,具有很高的实用价值。

The multi-station point cloud data collected by the terrestrial 3 D laser scanner is usually affected by the registration accuracy and the overlapping area, as it could generate the scattered cloud with large noise and redundancy after the multi-angle alignment, and then affected the surface reconstruction.Thus, an algorithm is proposed based on moving least squares resampling. Briefly, we calculate the k-neighborhood of the point cloud and establish the spatial topological relation of the scattered cloud, choose the appropriate basis function and weight function to establish the fitting function of local fitting region, then combine the VoxelGrid model to realize the resampling point cloud,and then use the Crust algorithm and the reverse engineering software Geomagic Studio to reconstruct the resampling point cloud respectively.Experimental results show that when the algorithm ensures local details clear on the basis of characteristics,the efficiency and surface smoothness of the reconstructed model of the surface are improved with a high practical value.