ITS_WS_ICCE is used to individual tree segmentation from airborne or UAV LiDAR point clouds combining the watershed and improved connection center evolution clustering. This code is used to individual tree segmentation for airborne or UAV LiDAR point clouds combining the watershed and connection center evolution clustering. This code allows batch processing.

Step 1: Run the individual_tree_seg_ws_li2012_batch.R Input: filterd point cloud(Classification:1[tree point],2[ground point]), las format Output: sengmented point clouds by the watershed algorithm, las format Attention: please set the resolution of CHM and DTM according to the densities of the point clouds. You can set the parameters of the watershed algorithm yourself.

Step 2: Run getbandwidth.py Input: sengmented point clouds by the watershed algorithm, las format Output: bandwidth value of each point clouds file, txt format

Step 3: Run its_ws_icce_batch.m Input: (1) sengmented point clouds by the watershed algorithm, las format; (2) bandwidth value of each point clouds file, txt format Output: (1) sengmented point clouds by the watershed+improved connection center evolution algorithm, las format; (2) tree structure metrics(Planar coordinates, tree height, crown diameter) Attention: You can set the appropriate parameter values yourself: Vr:the vertical distance correction factor(line 46); sigSq2: Gaussian kernel(line 47).

**Before you run the code the individual_tree_seg_ws_li2012_batch.R, you need to install the lidR library in R. If you encounter problems with the installation, we recommend changing the version of R (e.g. installing R 4.1.0).

**Before you run the code getbandwidth.py, you need to install the sklearn and laspy library in Python.This may take a very long time if the point cloud density is very high. Please be patient and wait for the program to finish running.

**We recommend using MATLAB R2018b and above, although lower versions will run as well.

Thanks to lidR(https://github.com/r-lidar/lidR) for providing the callable libraries and mparkan for providing the corresponding MATLAB functions(https://mparkan.github.io/Digital-Forestry-Toolbox/). Thanks to NEWFOR(https://www.newfor.net/) for providing free LiDAR point cloud data as the test data. Thanks to Xiurui Geng for the code of connection center evolution clustering(Geng, X., & Tang, H. (2020). Clustering by connection center evolution. Pattern Recognition, 98, 107063.https://doi.org/10.1016/j.patcog.2019.107063. ).

Last update: 2022/09/11

Please contact Yi Li ([email protected]) with any questions.