基于面向对象结合泰森多边形的杉木人工林冠幅提取研究OA

This study focused on a Cunninghamia lanceolata(Chinese Fir)plantation within a state-owned forest farm in Shunchang County,Fujian Province.Utilizing multispectral and point cloud data acquired by a DJI Matrice 300 RTK UAV equipped with LiDAR and multispectral sensors,alongside ground-truth survey data,we extracted the Canopy Height Model(CHM)and treetop positions.An object-based approach integrating the Thiessen polygon method was employed to delineate individual tree crowns in stands with high canopy closure,thereby providing an efficient method for crown extraction in a high-canopy-closure C.lanceolata plantation.The results indicated that by constructing the CHM from LiDAR point cloud data and applying a Local Maximum al-gorithm to the combined datasets,a treetop detection accuracy of 95.6％was achieved.The object-based Thiessen polygon segmentation method effectively delineated the crowns of C.lanceolata with an accuracy of 86％.The fusion of LiDAR-derived vertical structure information and multispectral texture features,coupled with Thiessen polygon segmentation,significantly enhanced the accuracy of crown delineation in high-density stands.This methodology provides a reliable technical pathway for dynamic biomass monitoring and carbon sink quantifica-tion in C.lanceolata plantations.