Abstract:Airborne laser scanning technique has became a reliable and accurate means for fast acquisition the three dimensional data of the landscape. The generation of complex 3D models is time-consuming work with many manual adjustments, therefore it is important to study how different types of data processing methods could ease the work, ideally, it is desirable to automatically generate the 3D-models using laser data, different types of objects, such as buildings, trees, roads, etc. However, there are still some challenging difficulties in the interpretation of the data acquired, and the modeling of different objects is not automatically. This paper focus on the extraction of trees and ground from airborne laser scanning points in forest area, we proposed a new method based on the multi-scale local density algorithm to segment trees and ground points, our algorithm can be divided into three steps, they are multi-scale neighbor searching, filtering, multi-scale segmentation.In the multi-scale neighborhood searching phase, we use the ANN method presented by Songrit Manee et al to search neighborhood of each reference point. We project the 3D scanned points to the horizontal plan to generate the 2D neighbors of each reference point. The 3D neighbors of each point are generated by directly searching points with 3D information. The number of points in each neighbor is determined by the radius used in searching phase and this radius is called scale. The multi-scale neighborhoods are generated by changing the radius increasingly.In the filtering step, statistical analysis of height values of points in each scale neighborhood is carried out. We find that there would be a break point in the height histogram of the neighbors if there is a tree above the ground and the height break points hold in some continuous scales. This characteristic is the basis for the filtering process. In the multi-scale segmentation step, we can use the height break characteristic to filter the ground points from the data set. We label the height of points below height correspond to the break point as the ground points and points above the height as unlabeled ones. This step is very useful to avoid the effect of ground points. Our process won't stop until all the unlabeled points are segmented to the ground data set and tree data set. Our test result has shown that our method has robust performance in the condition of complicated tree objects.

A Lidar Point Cloud Based Procedure for Vertical Canopy Structure Analysis And 3D Single Tree Modelling in Forest

Automated methods for measuring DBH and tree heights with a commercial scanning lidar

Canopy Structure Attributes Extraction from LiDAR Data Based on Tree Morphology and Crown Height Proportion

Delineation Of Individual Tree Crowns For Mobile Laser Scanning Data

A New Method for Segmenting Individual Trees from the Lidar Point Cloud

Automatic Forest Mapping at Individual Tree Levels from Terrestrial Laser Scanning Point Clouds with a Hierarchical Minimum Cut Method

Mobile LiDAR Scanning System Combined with Canopy Morphology Extracting Methods for Tree Crown Parameters Evaluation in Orchards

Development of a Precise Tree Structure from LiDAR Point Clouds

Segmentation Of Individual Trees Based On The 3-D Distribution Characteristics Of Point Cloud Data Obtained By Airborne Lidar

An Individual Tree Segmentation Method Based on Watershed Algorithm and Three-Dimensional Spatial Distribution Analysis From Airborne LiDAR Point Clouds

3D Segmentation of Individual Tree Canopy in Forest Nursery Based on Drone Image-matching Point Cloud

Segmentation Of Trees And Ground By Multi-Scale Local Density From Airborne Laser Scanning Data In Forest Areas

A New Method for 3D Individual Tree Extraction Using Multispectral Airborne LiDAR Point Clouds

Individual Tree Crown Delineation Using Localized Contour Tree Method and Airborne Lidar Data in Coniferous Forests

Segmentation of Individual Trees Based on a Point Cloud Clustering Method Using Airborne Lidar Data.

3D Modeling of Urban Tree Crown Volumes Using Multispectral LiDAR Data

Individual-Tree Segmentation from UAV–LiDAR Data Using a Region-Growing Segmentation and Supervoxel-Weighted Fuzzy Clustering Approach

Extracting Crown Morphology with a Low-Cost Mobile Lidar Scanning System in the Natural Environment

A bottom-up approach to segment individual deciduous trees using leaf-off lidar point cloud data

A 3D individual tree modeling technique based on terrestrial LiDAR point cloud data

A Novel Vegetation Point Cloud Density Tree-Segmentation Model for Overlapping Crowns Using UAV LiDAR