Abstract: To quickly and accurately extract the single tree point cloud from the stand point cloud obtained by the 3D terrestrial laser scanner, a tree segmentation algorithm based on DBSCAN (Density-Based Spatial Clustering of Application with Noise) was presented. Firstly, Gaussian filtering was used to denoise the stand point cloud, and the stand point cloud was vertically segmented on the basis of normalization, then the DBSCAN algorithm was used to cluster for each vertical segment, and the center point of each cluster in each vertical segment was calculated, the adjacent relationship between clusters was determined according to the distance between cluster center points, and the stem segment point cloud was matched by the adjacent relationship. Finally the RANSAC (Random Sample Consensus) algorithm was used to fit a straight line to the stem segment point cloud, the stem segmentation was performed by the distance between the point and the fitted line. In the stands with medium and high closures, the range of the F value of the presented algorithm was 0.88~0.99 and 0.72~0.74, respectively, and the F range of the tree segmentation algorithm based on distance discrimination was 0.84~0.90 and 0.73~0.79, respectively. The presented algorithm can effectively extract a single tree point cloud in the stand point cloud with different closures, especially in the stands with medium closure, and can perform the accurate tree segmentation from a stand point cloud.