基于点云的单木分割算法对比研究OA

Single tree segmentation based on point cloud is a prerequisite for extracting single tree parameters in forest resource surveys.Existing single tree segmentation algorithms have different segmentation effects for different parameters and research areas.In order to study the differences between different segmentation algorithms,this paper uses CHM-based watershed segmentation algorithm,layer-stack segmentation algorithm,top-down region growth algorithm and bottom-up region growth algorithm to segment forests with different complexity levels.The segmentation effects of different segmentation algorithms are compared and analyzed using Recall,Precision and F1 score as accuracy indicators.The results show that the top-down region growing algorithm has the best segmentation effect.Under complex forest conditions,each algorithm has over-segmentation and under-segmentation phenomena.