基于改进迭代粗分类的城区LiDAR点云滤波算法

Filtering Algorithm for LiDAR Point Clouds in Urban Areas Based on Improved Iterative Rough Classification

针对传统图像分割算法单次点云分类精度不高和数据格网内插精度损失等问题,提出一种改进的点云迭代粗分类方法。利用分割阈值对格网内插前的原始点云进行分类,避免内插造成的精度损失;对分类过程进行多次迭代,提升点云分类精度;滤波前剔除极低噪声点,避免其被选取为种子点而影响滤波精度。通过三组点云数据的滤波实验表明:本文滤波算法较传统的渐进三角网滤波算法精度高,且对于城镇、市中心城区等城区点云数据,均具有较好的适应性和良好的滤波效果。

Aiming at the problems of traditional image segmentation algorithm,such as single point cloud classification accuracy and data grid interpolation precision loss,an improved point cloud iterative rough classification method is proposed.The segmentation threshold is used to classify the original point cloud before the interpolation of the grid to avoid the loss of precision caused by interpolation.The iterative process is repeated several times to improve the classification accuracy of the point cloud;the extremely low noise point is removed before filtering to avoid it being selected for seed points.The filtering experiments of three sets of point cloud data show that the filtering algorithm of this paper has higher precision than the traditional progressive triangular mesh filtering algorithm,and it has good adaptability and good filtering effect for urban point cloud data in urban areas,downtown areas and other urban areas.