Individual Tree Segmentation Method Based on Mobile Backpack LiDAR Point Clouds.

Area of Science:

• Forestry science
• Remote sensing technology
• Geospatial data analysis

Background:

• Individual tree (IT) segmentation is vital for forest management tasks like inventory and biomass monitoring.
• Aerial laser scanning (ALS) provides good forest canopy data but struggles with under-canopy detail.
• Terrestrial laser scanners (TLS) are necessary for detailed ground-level data, especially in dense forests.

Purpose of the Study:

• To develop and evaluate an individual tree (IT) extraction method using terrestrial backpack LiDAR data.
• To assess the method's accuracy and efficiency for forest management applications.
• To demonstrate the utility of handheld TLS for improved forest data collection.

Main Methods:

• The study employed a novel IT extraction method utilizing DBSCAN clustering and cylinder voxelization on terrestrial backpack LiDAR point clouds.
• A sensibility assessment was incorporated to optimize input parameters for real-world data adaptation.
• The method focused on segmenting individual trees from dense forest point cloud data.

Main Results:

• The developed method achieved a high tree detection rate of approximately 90%.
• Commission and omission errors were low, resulting in an overall accuracy exceeding 93% for IT segmentation.
• The approach demonstrated effective segmentation of trees using handheld TLS data.

Conclusions:

• The proposed IT segmentation method using terrestrial laser scanning is accurate and efficient.
• Handheld TLS technology can significantly enhance data collection productivity for forest management.
• Accurate IT segmentation using TLS data supports improved forest inventory, biomass monitoring, and competition analysis.