期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2008
卷号：XXXVII-B8
页码：1397-1404
出版社：Copernicus Publications
摘要：This paper highlights several approaches to segment and reconstruct trees from LIDAR data and compares the results acquired both from first/last pulse and full waveform data. In a first step, we set up a conventional watershed based segmentation procedure, which robustly interpolates the CHM from the LIDAR data and finds possible stem positions of the tallest trees in the segments calculated from the local maxima of the CHM. Secondly, we combine this segmentation approach with a special stem detection method. Stem positions in the segments of the watershed segmentation are detected by hierarchically clustering points below the crown base height and reconstructing the stems with a robust RANSAC-based adjustment of the stem points. Finally, we implement a new 3D segmentation of single trees using the normalized cut method. This tackles the problem of how to segment small trees below the CHM. Experiments were conducted in the Bavarian Forest National Park with conventional first/last pulse data and full waveform LIDAR data. The first/last pulse data were collected in a flight with the Falcon II system from TopoSys in a leaf-on situation at a point density of 10 points/m 2 . Full waveform data were captured with the Riegl LMS Q-560 system at a point density of 25 points/m 2 (leaf-off and leaf-on) and at a point density of 10 points/m 2 (leaf-on). The study results prove that the new 3D segmentation approach is capable of detecting small trees in the lower forest layer. So far, this has been practically impossible when tree segmentation techniques based on the CHM were applied to LIDAR data. Compared to the standard watershed segmentation procedure, the combination of the stem detection method and the normalized cut segmentation leads to the best segmentation results and is superior in the best case by 12%. Moreover, the experiments show clearly that the usage of full waveform data is superior to first/last pulse data
