大范围密集点云渐进加密三角网滤波改进算法

An improved progressive TIN densification filtering algorithm for large scale dense point cloud

针对传统渐进加密三角网滤波算法处理大范围密集点云效率较低,同时低点噪声影响滤波精度的问题,该文提出了一种渐进加密三角网滤波改进算法。在对大范围密集点云进行抽稀、低点噪声过滤以及分块处理基础上,采用并行方式对分块点云进行滤波,滤波过程中通过标记三角网状态以及限制最短边长加快迭代速度,最后将各分块滤波结果合并得到完整的地面点云。经实验表明,该算法能提升点云滤波效率,且点云范围越大、点密度越高,效率提升越明显;通过低点噪声过滤,减小了滤波结果的Ⅰ类误差。通过对不同场景的点云进行滤波处理,验证了算法的适应性。

To address the low efficiency of traditional progressive TIN densification filtering algorithm in processing large-scale dense point clouds and the impact of low-point noise on filtering accuracy,this article proposed an improved progressive TIN densification filtering algorithm.On the basis of down sampling,low-point noise filtering,and block processing for dense point clouds in a large area,parallel filtering was adopted for the divided point clouds.During the filtering process,the triangular mesh status was marked and the shortest edge length was limited to accelerate the iteration speed.Finally,the filtered results from each block were merged to obtain a complete ground point cloud.The experimental results demonstrated that this algorithm can improve the efficiency of point cloud filtering,and the larger the point cloud range and the higher the point density,the more significant the efficiency improvement.By filtering out low-point noise,the type I error of the filtering results was reduced.In addition,the adaptability of the algorithm was verified by filtering point cloud from different scenes.