Single tree segmentation and crown information extraction based on PSO adaptive marker watershed algorithm

Individual tree extraction is challenging from LiDAR point cloud data of artificial ginkgo forests with high canopy density. To solve this problem, this paper proposes an adaptive marker-controlled watershed algorithm based on Particle Swarm Optimization (PSO). The algorithm dynamically optimizes key parameters to reduce reliance on manual parameter tuning and achieve high-precision individual tree segmentation. Experimental results show that the matching accuracy of the proposed algorithm with field survey data is 78%, higher than 75% of the traditional algorithm. The inversion accuracy of tree height and crown width reaches 91.3% and 72%, respectively. The proposed algorithm effectively improves the application performance of LiDAR in high-density artificial forest land and provides reliable technical support for refined forest resource investigation.