基于最小二乘密度聚类的城市点云去噪算法

点云作为一种简便的三维表达方式,已经被大量应用在城市三维数字化中,但是城市对象的复杂多变,导致城市点云相较于其他场景点云,其较为复杂,去噪难度更高,去噪精度要求更高。为了解决城市点云的去噪问题,本文从高维特征密度空间出发,采用最小二乘密度聚类约束,遵循标准阈值原则设计了一种新的算法。本算法先构建高维特征密度空间,再用最小二乘算法求解各维度密度拟合曲线,最后根据标准阈值原则提取各维度合限点集的交集,即为目标点集。实验表明:本文算法针对城市场景中的点云具有较高的精度与较好的剔除效果,满足城市点云去噪任务的要求,达到了预期的效果。

Urban point cloud denoising algorithm based on least square density clustering

As a simple three-dimensional expression,point cloud has been widely applied in the three-dimensional digitalization of cities.However,the complexity and variability of urban objects lead to the fact that compared with other cloud sites;point cloud is more complex,with higher difficulty in denoising and higher requirement for denoising accuracy.In order to solve the problem of urban point cloud denoising,this paper designs a new algorithm based on high-dimensional feature density space,using least-squares density clustering constraint and following the standard threshold principle.This algorithm first constructs the high-dimensional characteristic density space,then solves the density fitting curve of each dimension with the least square algorithm,and finally extracts the intersection of the convergence point set of each dimension according to the standard threshold principle,that is,the target point set.The experiment shows that the algorithm in this paper has high precision and good removal effect for point cloud in urban scene,which meets the requirements of urban point cloud denoising task and achieves the expected effect.