Adaptive Mean Shift-Based Identiﬁcation of Individual Trees Using Airborne LiDAR Data

Xingbo Hu,Wei Chen,Weiyang Xu

doi:10.3390/rs9020148

Abstract

Identifying individual trees and delineating their canopy structures from the forest point clouddataacquiredbyanairborneLiDAR(LightDetectionAndRanging)hassigniﬁcantimplications in forestry inventory. Once accurately identiﬁed, tree structural attributes such as tree height, crown diameter, canopy based height and diameter at breast height can be derived. This paper focuses on a novel computationally efﬁcient method to adaptively calibrate the kernel bandwidth of a computational scheme based on mean shift—a non-parametric probability density-based clustering technique—to segment the 3D (three-dimensional) forest point clouds and identify individual tree crowns. The basic concept of this method is to partition the 3D space over each test plot into small vertical units (irregular columns containing 3D spatial features from one or more trees) ﬁrst, by using a ﬁxed bandwidth mean shift procedure and a small square grouping technique, and then rough estimation of crown sizes for distinct trees within a unit, based on an original 2D (two-dimensional) incremental grid projection technique, is applied to provide a basis for dynamical calibration of the kernel bandwidth for an adaptive mean shift procedure performed in each partition. The adaptive mean shift-based scheme, which incorporates our proposed bandwidth calibration method, is validated on 10 test plots of a dense, multi-layered evergreen broad-leaved forest located in South China. Experimental results reveal that this approach can work effectively and when compared to the conventional point-based approaches (e.g., region growing, k-means clustering, ﬁxed bandwidth or multi-scale mean shift), its accuracies are relatively high: it detects 86 percent of the trees (“recall”) and 92 percent of the identiﬁed trees are correct (“precision”), showing good potential for use in the area of forest inventory.

Highlights

In the past two decades, efforts have been made to decrease forest inventory costs by using data obtained from remote sensing technology
We propose an adaptive mean shift-based clustering scheme dedicated to the segmentation of 3D forest point clouds and identification of individual tree crowns
We develop an adaptive mean shift scheme to segment the forest point clouds into 3D clusters and properly identify and extract individual tree crowns over a multi-layered forest in China

Summary

Introduction

In the past two decades, efforts have been made to decrease forest inventory costs by using data obtained from remote sensing technology. One of the most popular clustering methods is the k-means algorithm, which aims to partition n observations into k clusters where each observation belongs to the cluster with the nearest mean by trying to minimize the overall sum of Euclidean distances of the points in feature space to their cluster centroids [13,14,15] Most of these methods fall into one of the two categories: raster-based approaches or point-based approaches. Point-based approaches work directly on the original airborne laser scanning (ALS) data (namely the “point cloud”, a discrete set of point locations) rather than rasterized images This category of methods is preferable because it provides a promising way of producing relatively high accuracy by detecting smaller trees in the understory than CHM-based methods

Results

Discussion

Conclusion