Abstract: 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 algorithm 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 quantification in C. lanceolata plantations.