基于体素化网格下采样的点云简化算法研究

针对三维点云数据冗余量大、重建时间长、效率低等问题，提出一种基于体素化网格下采样的点云简化算法。该算法首先求出点云数据集的最小三维长方体包围盒，把点云数据划分进三维体素栅格中去;其次计算点云的k邻域，进行曲面法向量估计；然后，在三维体素栅格中选择满足要求的数据点，实现点云下采样；最后，调用Power Crust对下采样点云数据进行曲面重建，在三维可视化类库Visualization Toolkit(VTK)进行显示。实验结果表明，该算法能够加快三维点云数据的重建速度，较好地保持了点云特征，提高曲面重建的效率和鲁棒性，适合实时处理。

Research on simplification algorithm of point cloud based on voxel grid

Focus on the issue that high data redundancy, long reconstruction time and low efficiency exist in three-dimensional point cloud data, a simplification algorithm of point cloud based on 3D voxel grid is proposed. Firstly, the minimum three-dimensional rectangular bounding box is calculated, and point cloud are divided into the 3D voxel grid. Secondly, the k-nearest neighbors of point cloud is searched, and the surface normal vector is estimated. And then, sample points by using uniform 3D voxel grids to select the points that satisfy the requirement. Finally, using Power Crust algorithm reconstructing surface, and displaying in the three-dimensional visualization library VTK(Visualization Toolkit). The experimental results show that the proposed method can speed up the reconstructing rate of three-dimensional point cloud, retain geometric characteristics of original dates, improve the surface reconstructing efficiency and robustness, and be able to real time processing.