Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Ground Profile Recovery from Aerial 3D LiDAR-based Maps

Adelya Sabirova, Maksim Rassabin, Roman Fedorenko, Ilya Afanasyev

TL;DR

The paper addresses ground detection in UAV-derived LiDAR point clouds over forests to reconstruct terrestrial relief and generate DTMs. It proposes an offline workflow centered on Cloth Simulation Filtering (CSF) with prior outlier removal and point cloud normalization to separate ground from vegetation. Demonstrated on a real DJI M600Pro/VLP16 dataset in a mixed forest, the approach yields about 20% ground points and stable ground profiles despite ~40 m relief changes, illustrating robustness. The work integrates CSF within CloudCompare and a ROS-based data flow for terrain reconstruction, with metrological verification identified as future work.

Abstract

The paper presents the study and implementation of the ground detection methodology with filtration and removal of forest points from LiDAR-based 3D point cloud using the Cloth Simulation Filtering (CSF) algorithm. The methodology allows to recover a terrestrial relief and create a landscape map of a forestry region. As the proof-of-concept, we provided the outdoor flight experiment, launching a hexacopter under a mixed forestry region with sharp ground changes nearby Innopolis city (Russia), which demonstrated the encouraging results for both ground detection and methodology robustness.

Ground Profile Recovery from Aerial 3D LiDAR-based Maps

TL;DR

The paper addresses ground detection in UAV-derived LiDAR point clouds over forests to reconstruct terrestrial relief and generate DTMs. It proposes an offline workflow centered on Cloth Simulation Filtering (CSF) with prior outlier removal and point cloud normalization to separate ground from vegetation. Demonstrated on a real DJI M600Pro/VLP16 dataset in a mixed forest, the approach yields about 20% ground points and stable ground profiles despite ~40 m relief changes, illustrating robustness. The work integrates CSF within CloudCompare and a ROS-based data flow for terrain reconstruction, with metrological verification identified as future work.

Abstract

The paper presents the study and implementation of the ground detection methodology with filtration and removal of forest points from LiDAR-based 3D point cloud using the Cloth Simulation Filtering (CSF) algorithm. The methodology allows to recover a terrestrial relief and create a landscape map of a forestry region. As the proof-of-concept, we provided the outdoor flight experiment, launching a hexacopter under a mixed forestry region with sharp ground changes nearby Innopolis city (Russia), which demonstrated the encouraging results for both ground detection and methodology robustness.

Paper Structure

This paper contains 11 sections, 15 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. The first step: 3D point cloud generation
  4. The second step: Outlier detection and removal
  5. The third step: Cloth Simulation Filtering (CSF) implementation
  6. The forth step: Point cloud normalization and Building terrestrial relief
  7. System Setup
  8. Aerial mapping system
  9. Sensor data processing
  10. Experimental results
  11. Conclusions

Figures (15)

  • Figure 1: The block-scheme of the proposed methodology for ground detection using filtration and removal of forest points from airborne 3D LiDAR-related point cloud based on the Cloth Simulation Filtering (CSF) algorithm
  • Figure 2: The illustration of the Cloth Simulation Filtering (CSF) algorithm. The original point cloud is turned upside down, and then a simulated fabric falls on the inverted surface from above, dividing the point clouds into ground and non-ground parts. © Courtesy of Zhang, et al. zhang2016
  • Figure 3: The experimental aerial mapping system based on the DJI M600Pro Hexacopter with the stabilized Velodyne VLP16 LiDAR
  • Figure 4: DJI GroundControl app with RViz GUI
  • Figure 5: Rosbag processing dataflow for 3d point cloud map construction by processing a sequence of sensor data from the onboard Hexacopter dataset
  • ...and 10 more figures