GCMex file (became invalide for matlab 2020a and later) for windows64 is updated
Unsupervised tree leaf-wood classification from point cloud data (for plot-scale data and single trees)
--- (An upgraded version is coming soon!) ---
There are many ways to use this tool.
(a) if you have Matlab installed:
Option 1. Call the entry level funtion "RecursiveSegmentation_release.m" as:
“[BiLabel, BiLabel_Regu] = RecursiveSegmentation_release(points, ft_threshold, paral, plot);”
Inputs:
% points: this is your nx3 data matrix.
% ft_threshold: feature threshold. suggest using 0.125 or so
% paral: if shut down parallel pool after segmentation (1 or other).
% plot: if plot results in the end (1 or other)
Outputs:
% BiLabel: point label without regularization
% BiLabel_Regu: point label with regularization
Option 2. Type "LeWoS_RS" in Matlab workspace. This will open up an interface by calling the classdef "LeWoS_RS.m". This classdef file defines the interface.
Option 3. Drag "LeWoS.mlappinstall" into Matlab workspace. This will install a Matlab App for you.
(b) if you don't have Matlab installed, and don't want to install it:
Run "LeWoS_installer.exe" for win64. If you need an excutable for other systems (Linux and Mac), please contact me.
(PS: Matlab Runtime 2019b (freely available at https://se.mathworks.com/products/compiler/matlab-runtime.html) is required. You can either install it in advance or do it during the installation of LeWoS.)
--------------------------
*Note that if you load an ascii point cloud with the interface, only space delimiter is supported (without header). Currently, these formats are supported: .las; .mat; .xyz; .txt; .ply; .pcd (recommend to use more generic formats for point clouds, such as las, ply, and pcd)
*This method does not implement any post-processing filters. Users can design and apply post-processing steps to [potentially] further improve the results.
Plot-level separation
Inside a crown
Very thin branches are difficult to detect
This repo contains code from Loic Landrieu's repo on point-cloud-regularization (https://github.com/loicland/point-cloud-regularization), and Inverse Tampere's repo on TreeQSM (https://github.com/InverseTampere/TreeQSM).
@article{doi:10.1111/2041-210X.13342,
author = {Wang, Di and Momo Takoudjou, Stéphane and Casella, Eric},
title = {LeWoS: A universal leaf-wood classification method to facilitate the 3D modelling of large tropical trees using terrestrial LiDAR},
journal = {Methods in Ecology and Evolution},
volume = {11},
number = {3},
pages = {376-389},
keywords = {biomass estimation, component separation, leaf-wood separation, LeWoS, point cloud, terrestrial LiDAR, TreeQSM, tropical forest trees},
doi = {10.1111/2041-210X.13342},
url = {https://besjournals.onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13342 },
year = {2020}
}
(Current code is a slightly updated version of the one used in publication. With current one, the results are further improved a bit. e.g., 0.925 ± 0.035 vs 0.91 ± 0.03 in the paper.)
Di Wang
di-wang@foxmail.com