基于多尺度虚拟格网的点云自适应滤波算法

Adaptive Filtering Algorithm of Point Cloud Based on Multi-Scale Virtual Grid

点云滤波是LiDAR点云数据处理中的重要步骤,针对传统曲面拟合滤波算法参数求解不合理,滤波阈值单一且自适应较差的问题,提出一种改进自适应阈值滤波算法。首先对预处理之后的点云数据引入虚拟格网进行分割并依据邻域格网选取种子点,然后采用混合最小二乘法对曲面拟合参数进行解算,计算真实高程与拟合高程的差值并结合k-means聚类与正态分布确定自适应阈值,最后采用多级滤波的策略逐级改变虚拟网格大小实现点云的高精度滤波。将该算法与其它经典滤波算法进行对比分析,试验结果表明:本文算法在不同场景下有良好的滤波精度,且稳定性较好。

Point cloud filtering is an important step in LiDAR point cloud data processing. Aiming at the prob-lems of unreasonable parameter solving, single filtering threshold and poor adaptability of tradi-tional surface fitting filtering algorithm, an improved adaptive threshold filtering algorithm was proposed. Firstly, the pre-processed point cloud data is introduced into the virtual grid for segmen-tation, and the seed points are selected according to the neighborhood grid. Then, the mixed least squares method is used to solve the fitting parameters of the surface, and the difference between the real elevation and the fitted elevation is calculated, and the adaptive threshold is determined by combining K-means clustering and normal distribution. Finally, the multi-stage filtering strategy is used to change the size of virtual grid step by step to achieve high precision filtering of point clouds. The proposed algorithm is compared with other classical filtering algorithms, the experi-mental results show that the proposed algorithm has good filtering accuracy and good stability in different scenes.