An improved filter of progressive TIN densification for LiDAR point cloud data

Abstract

Based on the classic filter of progressive triangulated irregular network (TIN) densification, an improved filter is proposed in this paper. In this method, we divide ground points into grids with certain size and select the lowest points in the grids to reconstruct TIN in the process of iteration. Compared with the classic filter of progressive TIN densification (PTD), the improved method can filter out attached objects, avoid the interference of low objects and obtain relatively smooth bare-earth. In addition, this proposed filter can reduce memory requirements and be more efficient in processing huge data volume. The experimental results show that the filtering accuracy and efficiency of this method is higher than that of the PTD method.Airborne light detection and ranging (LiDAR) technology can obtain high-precision and high-density point cloud data of surface [1] , which is a fast growing geodetic imaging technology. LiDAR point cloud has been widely used in many fields, such as cultural heritage documentation, reverse engineering, three-dimensional (3D) object reconstruction, and digital elevation model (DEM) generation. For the production of DEM, filtering must be executed to separate ground points and non-ground points. The existing filtering methods can be classified into four categories: morphological-based filter [2][3][4] , interpolation-based filter [5][6][7][8] , slope-based filter [9][10][11] and segmentation/cluster-based filter [12][13][14] . Thereinto, the filter of progressive triangulated irregular network (TIN) densification belongs to the interpolation-based filter [8] . By analyzing several existing filters, Sitbole and Vosselman [15] pointed out that the filter of progressive TIN densification (PTD) proposed by Axelsson [6] performed best, but it could not remove attached objects [16] , such as bridge and ramp. Zhang and Lin [8] proposed a developed PTD using a point cloud segmentation method, namely segmentation using smoothness constraint (SUSC). Both the PTD and the developed PTD are invalid for attached objects and require lots of memory to construct TIN. Insufficient memory poses great challenges to the TIN-based filters, due to huge data volume and complexity of objects.In order to remove attached objects and reduce memory requirements, we propose an improved filter of

An Improved Progressive TIN Densification Filtering Method Considering the Density and Standard Variance of Point Clouds

“…Referring to Refs. [26,27], the distance threshold, angle threshold, L 1 and L 2 are 1.4 m, 6 • , 5 m and 2 m, respectively.…”

“…The experimental results in the paper show that surface-based filtering algorithms perform better than other filtering algorithms. According to the previous study, the surface-based filtering algorithms are divided into three sub-categories: morphology-based filtering [19][20][21][22], iterative-interpolation-based filtering [23,24] and progressive-densification-based filters [25][26][27][28]. Among the surface-based filtering algorithms, the progressive TIN (triangular irregular network) densification (PTD) algorithm proposed by Axelsson [25] is most widely used because of its robustness and effectiveness in the separation of ground points and non-ground points.…”

“…The experimental results show this method can remove the lower parts of large objects attached on the ground surface. Wang et al [27] revised the iterative densification process of the PTD algorithm. Specifically, they first divided the ground points into grids; then, selected a lowest point as the ground point and classified other points as non-ground points to be judged by the thresholds again within each grid; and densified the TIN until the ratio of the difference calculated by the last two new ground points added into the TIN to the total points is less than a given threshold during two consecutive iterations.…”

A Point Cloud Filtering Approach to Generating DTMs for Steep Mountainous Areas and Adjacent Residential Areas

“…An optimal selection method based on confidence interval estimation theory is then applied to eliminate the erroneous points among the seed points. The memory-efficient TIN densification strategy proposed in our previous work [49] is adopted to improve the processing efficiency for point clouds with high density. The experiments on the two datasets demonstrate that the proposed method can achieve relatively high accuracy for both residential and mountainous areas.…”

“…A memory-efficient TIN densification strategy for high-density point clouds is adopted to address the aforementioned problems. This strategy was proposed in our previous work [49] and can be summarized in the following steps:…”