Lidar-Based Individual Tree Segmentation: A Novel Algorithm and Accuracy Assessment

In summary, the algorithm presented in this paper utilizes the 3D structure of lidar point clouds to segment individual trees in forested areas. The algorithm incorporates a novel quantification of tree shape and avoids the common problem of determining the size of a search window for tree top detection. The accuracy of the algorithm depends on the spacing threshold, which can be adapted for different forest types. The algorithm is computation-intensive but can directly measure important tree parameters from the points contained in each tree. The F-score is recommended for accuracy assessment, which should be further investigated and standardized for consistent comparison across literature. Future research should also comprehensively evaluate the accuracy of tree measurements based on different tree segmentation methods.